---
title: "Oct 24: Phone security and owner control, and more"
show: "RobotPaper's daily HN podcast"
canonical: "https://podcasts.robotpaper.ai/hn/ep/2026-10-24/"
date: "2026-10-24T02:23:08+00:00"
duration: "28:03"
duration_seconds: 1683
audio: "https://podcasts.robotpaper.ai/hn/ep/2026-10-24/2026-10-24-hn.mp3"
chapters: "https://podcasts.robotpaper.ai/hn/ep/2026-10-24/chapters.json"
transcript: "https://podcasts.robotpaper.ai/hn/ep/2026-10-24.txt"
feed: "https://podcasts.robotpaper.ai/hn/feed.xml"
language: "en"
author: "Roy Osherove"
ai_generated_voices: true
---

# Oct 24: Phone security and owner control, and more

RobotPaper's daily HN podcast · Oct 24, 2026 · 28 min

- Listen: [MP3, 27 MB](https://podcasts.robotpaper.ai/hn/ep/2026-10-24/2026-10-24-hn.mp3)
- Subscribe via RSS: https://podcasts.robotpaper.ai/hn/feed.xml (paste into any podcast app)
- Web page: https://podcasts.robotpaper.ai/hn/ep/2026-10-24/
- Chapters (JSON): https://podcasts.robotpaper.ai/hn/ep/2026-10-24/chapters.json
- Transcript (text): https://podcasts.robotpaper.ai/hn/ep/2026-10-24.txt
- Show: [RobotPaper's daily HN podcast](https://podcasts.robotpaper.ai/hn/) (RSS: https://podcasts.robotpaper.ai/hn/feed.xml)

## Chapters

- `00:00:00` Intro
- `00:00:35` Phone owner control
- `00:04:23` Kimi K3.1 Pro
- `00:06:53` Disney pricing and ads
- `00:09:17` Utility ownership gap
- `00:11:47` Natural language code
- `00:13:51` Apple Music iPod page
- `00:15:35` Kushner phone claims
- `00:17:19` AI futures in history
- `00:19:39` Cyber disclosure
- `00:22:02` AI carbon accounting
- `00:23:52` Google data and ads
- `00:24:28` Nitter fork
- `00:25:02` Benchmark removal
- `00:25:40` Border heat ray tests
- `00:26:14` Sweden AI access rules
- `00:26:54` Job loss and illness
- `00:27:31` Outro

## Show notes

### Phone owner control (2 stories)

First up, from the Calif blog: researchers demonstrate iPhone hacks at Hexacon.

- [Today, we hacked what we'd set out to hack: the iPhone](https://blog.calif.io/p/iphone) · [HN discussion](https://news.ycombinator.com/item?id=50211474) \| quoted: [reacharavindh](https://news.ycombinator.com/user?id=reacharavindh), [traceroute66](https://news.ycombinator.com/user?id=traceroute66), [jeroenhd](https://news.ycombinator.com/user?id=jeroenhd), [Zak](https://news.ycombinator.com/user?id=Zak)
- [Your Phone Is About to Stop Being Yours: 70 Days Until Lockdown](https://keepandroidopen.org/?) · [HN discussion](https://news.ycombinator.com/item?id=50208558) \| quoted: [rjh29](https://news.ycombinator.com/user?id=rjh29), [FiloSottile](https://news.ycombinator.com/user?id=FiloSottile), [Mr\_Modulo](https://news.ycombinator.com/user?id=Mr_Modulo), [catgirlinspace](https://news.ycombinator.com/user?id=catgirlinspace), [svat](https://news.ycombinator.com/user?id=svat)
- [Kimi K3.1 Pro and K3.1 Pro Highspeed](https://www.kimi.ai/blog/kimi-k3-1-pro) · [HN discussion](https://news.ycombinator.com/item?id=50212065) \| quoted: [rudolftheone](https://news.ycombinator.com/user?id=rudolftheone), [NitpickLawyer](https://news.ycombinator.com/user?id=NitpickLawyer), [stalfie](https://news.ycombinator.com/user?id=stalfie), [johnmlussier](https://news.ycombinator.com/user?id=johnmlussier), [SwellJoe](https://news.ycombinator.com/user?id=SwellJoe), [pjerem](https://news.ycombinator.com/user?id=pjerem), [besterman23](https://news.ycombinator.com/user?id=besterman23), [ByteAtATime](https://news.ycombinator.com/user?id=ByteAtATime)
- [Disney Raising Prices Days After Confirming Ads Will Now Show Up on Every Tier](https://www.tvfandomlounge.com/disney-plus-price-hike-ads-premium-tier-nielsen-emmys-2026/) · [HN discussion](https://news.ycombinator.com/item?id=50205934) \| quoted: [jerf](https://news.ycombinator.com/user?id=jerf), [Macha](https://news.ycombinator.com/user?id=Macha), [saalweachter](https://news.ycombinator.com/user?id=saalweachter), [ricardobeat](https://news.ycombinator.com/user?id=ricardobeat), [bofaGuy](https://news.ycombinator.com/user?id=bofaGuy)
- [Investor-owned utilities charge 23% more for electricity than public providers](https://www.kadoa.com/energy-prices/insights/investor-owned-utilities) · [HN discussion](https://news.ycombinator.com/item?id=50208837) \| quoted: [solfox](https://news.ycombinator.com/user?id=solfox), [baron816](https://news.ycombinator.com/user?id=baron816), [kccqzy](https://news.ycombinator.com/user?id=kccqzy), [ianm218](https://news.ycombinator.com/user?id=ianm218), [Hnrobert42](https://news.ycombinator.com/user?id=Hnrobert42)
- [The foolishness of "natural language programming" (1978)](https://www.cs.utexas.edu/~EWD/transcriptions/EWD06xx/EWD667.html) · [HN discussion](https://news.ycombinator.com/item?id=50203161) \| quoted: [benrutter](https://news.ycombinator.com/user?id=benrutter), [zby](https://news.ycombinator.com/user?id=zby), [nickm12](https://news.ycombinator.com/user?id=nickm12), [glimshe](https://news.ycombinator.com/user?id=glimshe), [GianFabien](https://news.ycombinator.com/user?id=GianFabien)
- [Apple Music 404 Not Found](https://music.apple.com/this-url-doesnt-exist) · [HN discussion](https://news.ycombinator.com/item?id=50207910) \| quoted: [sherburt3](https://news.ycombinator.com/user?id=sherburt3), [98codes](https://news.ycombinator.com/user?id=98codes), [ashleyn](https://news.ycombinator.com/user?id=ashleyn), [eddythompson80](https://news.ycombinator.com/user?id=eddythompson80), [cosmic\_cheese](https://news.ycombinator.com/user?id=cosmic_cheese), [swiftcoder](https://news.ycombinator.com/user?id=swiftcoder)
- [Dutch government had hacked into Jared Kushner's phone](https://twitter.com/jaredkushner/status/2113376217415958767) · [HN discussion](https://news.ycombinator.com/item?id=50206914) \| quoted: [TrackerFF](https://news.ycombinator.com/user?id=TrackerFF), [lukewarm707](https://news.ycombinator.com/user?id=lukewarm707), [klappgrt](https://news.ycombinator.com/user?id=klappgrt), [pacifika](https://news.ycombinator.com/user?id=pacifika)

### AI futures in history (2 stories)

Next, Jim Nielsen's blog examines retirement, nuclear fears, and predictions of an AI utopia.

- [No Need for Retirement When Utopia Is Around the Corner](https://blog.jim-nielsen.com/2026/skip-retirement-for-utopia/) · [HN discussion](https://news.ycombinator.com/item?id=50207050) \| quoted: [paxy](https://news.ycombinator.com/user?id=paxy), [datsci\_est\_2015](https://news.ycombinator.com/user?id=datsci_est_2015), [epiccoleman](https://news.ycombinator.com/user?id=epiccoleman)
- [We are all living inside of an obscure sci-fi film from 1970](https://thanksforthatjim.bearblog.dev/we-are-all-living-inside-of-an-obscure-sci-fi-film-from-1970/) · [HN discussion](https://news.ycombinator.com/item?id=50214323) \| quoted: [Quartzium](https://news.ycombinator.com/user?id=Quartzium), [JoeAltmaier](https://news.ycombinator.com/user?id=JoeAltmaier), [vapel](https://news.ycombinator.com/user?id=vapel)

### Cyber disclosure (2 stories)

From the Silcock blog: is this the new cyber normal?

- [Is this the new cyber normal?](https://drew.silcock.dev/blog/is-this-the-new-cyber-normal/) · [HN discussion](https://news.ycombinator.com/item?id=50207115) \| quoted: [thaumaturgy](https://news.ycombinator.com/user?id=thaumaturgy), [aftbit](https://news.ycombinator.com/user?id=aftbit), [Arainach](https://news.ycombinator.com/user?id=Arainach)
- [eMudhra ignored Mozilla Root investigation deadline; blames social media](https://bugzilla.mozilla.org/show_bug.cgi?id=2083449) · [HN discussion](https://news.ycombinator.com/item?id=50210067) \| quoted: [jwineinger](https://news.ycombinator.com/user?id=jwineinger), [tptacek](https://news.ycombinator.com/user?id=tptacek), [IcePenguin](https://news.ycombinator.com/user?id=IcePenguin), [ForHackernews](https://news.ycombinator.com/user?id=ForHackernews), [wolvoleo](https://news.ycombinator.com/user?id=wolvoleo)
- [The same AI request emits 8.5 times more CO2 in Texas than in Sweden](https://berget.ai/en/blog/per-request-co2-press-release) · [HN discussion](https://news.ycombinator.com/item?id=50205229) \| quoted: [blaze33](https://news.ycombinator.com/user?id=blaze33), [alpineman](https://news.ycombinator.com/user?id=alpineman), [lawn](https://news.ycombinator.com/user?id=lawn), [mepiethree](https://news.ycombinator.com/user?id=mepiethree), [mkgiga](https://news.ycombinator.com/user?id=mkgiga), [kleiba2](https://news.ycombinator.com/user?id=kleiba2)
- [Now Google says it might use your Gmail and Google Photos data to target ads](https://www.theverge.com/tech/1017600/google-personal-intelligence-privacy-policy-ads-update) · [HN discussion](https://news.ycombinator.com/item?id=50214218)
- [Shitter – A Nitter Fork](https://codeberg.org/shitter/shitter) · [HN discussion](https://news.ycombinator.com/item?id=50208343)
- [Remove pgrust: the entry is cheating (per-query kernels and result caching)](https://github.com/ClickHouse/ClickBench/pull/2165) · [HN discussion](https://news.ycombinator.com/item?id=50205816)
- [Pentagon grants DHS permission to test heat ray weapon for use at border](https://www.cnn.com/2026/10/21/politics/heat-ray-weapon-dhs-pentagon) · [HN discussion](https://news.ycombinator.com/item?id=50210075)
- [Swedish government offices to ban ChatGPT and Claude](https://www.sverigesradio.se/artikel/chat-gpt-och-claude-forbjuder-regeringskansliet) · [HN discussion](https://news.ycombinator.com/item?id=50207405)
- [Job loss and cortisone mania](https://tilschuenemann.de/job-loss-cortisone-mania) · [HN discussion](https://news.ycombinator.com/item?id=50206840)

### Music

"Sneaky Snitch" Kevin MacLeod (incompetech.com)\
Licensed under Creative Commons: By Attribution 4.0\
http://creativecommons.org/licenses/by/4.0/

The voices you're hearing are AI-generated; the stories come from public sources linked in the show notes.\
Music: Eleven Music.\
An independent project; not affiliated with or endorsed by Amazon Web Services, Y Combinator, Hacker News or any vendor or news outlet mentioned. Opinions are the hosts' own.\
Produced by Roy Osherove.

## Transcript

\[Linus\] Good morning, it's Saturday, October twenty-fourth, and this is RobotPaper's daily HN podcast. I'm Linus. The voices you're hearing are AI-generated; the stories come from news dot Y Combinator dot com.

\[Dennis\] I'm Dennis, and today, iPhone hacking demonstrations renew the debate over phone control.

\[Ada\] I'm Ada, and we also have Kimi's new models and Disney's pricing changes.

\[Grace\] I'm Grace, and as always, this podcast is generated by robotpaper dot A I, by Roy Osherove. Let's go.

\[Ken\] First up, from the Calif blog: researchers demonstrate iPhone hacks at Hexacon. The demonstrations have prompted a debate about protecting phones while allowing their owners to control the software. Two stories today touch on this question, with Keep Android Open also challenging Google's developer verification requirements. Calif's short announcement says the demonstrations were presented in Paris and promises further releases in due course. It labels one demonstration as a one-click attack involving Signal and also refers to iMessage. The full article could not be read, so the available announcement does not establish the complete exploit chain or which protections defeat it. Those questions remain distinct from what people infer from the demonstrations. On Keep Android Open, the companion story argues that Android's planned developer registration requirements will restrict independent software distribution. The campaign says registration involves a fee, identification, agreement to Google's terms, and evidence of control over signing keys. It objects to applying those requirements beyond the Play Store, including software distributed by hobbyists and alternative stores. The page describes changes beginning in twenty twenty-seven. Its opening claims there is no opt-out, but its own account also describes an advanced installation process through developer mode, a restart, and a twenty-four-hour waiting period, with temporary or indefinite permission afterward. That exception matters to the scope of the claim. The campaign argues that the process still puts too much control in Google's hands because it depends on Play Services. These stories seem to be drawing attention because the ability to modify a phone, the ability to distribute an app, and protection against an attacker all intersect with decisions made by the platform owner.

\[Ada\] And the comments are divided between support for owner control and concern about the consequences of easier access. In the Calif discussion, reacharavindh argues that jailbreaking still leaves an owner dependent on Apple's online services and on maintaining changes across updates. traceroute66 takes the other side, pointing to banking, authentication, email, and family photographs on a modern phone as reasons to favor stronger restrictions. Zak separates the two issues: being allowed to change one's own device is desirable, while letting someone take it over through a document is not. jeroenhd asks whether Lockdown Mode prevents the demonstrated attacks. Replies disagree about attachment handling and how restrictive that mode is, so the discussion does not establish a verified answer for these demonstrations. In the Android discussion, rjh29 challenges the campaign's no-opt-out wording by pointing to developer mode and the waiting period. FiloSottile argues that the delay addresses a specific threat: scammers coaching someone through installation of an app that takes over the phone. That commenter considers the waiting period a reasonable compromise between freedom and protection. Mr\_Modulo disagrees, arguing that security does not justify giving Google that degree of control over installation. Another correction comes from catgirlinspace, who says the waiting period is a one-time process, rather than a delay on every app installation. svat frames the change as a restriction on anonymous developers reaching a broad audience, rather than a complete end to installing independently distributed software. Several commenters also raise compatibility concerns for alternative systems and applications that reject developer mode. The disagreement is therefore about both the security benefit of added friction and who gets to decide the conditions under which independent software can run.

\[Dennis\] From the Kimi blog, Kimi K three point one Pro and Pro Highspeed. Kimi's new model series combines a million-token context window with claimed improvements in coding, three-dimensional design and financial research. The company describes Pro as a post-trained version of K three, using substantially more post-training compute. Both versions are available through Kimi's applications, coding tools and API, while the full model weights are promised by November fourth. Highspeed is advertised as two to three times faster. Kimi reports sixty-five point seven percent on Terminal-Bench four point zero, a coding-agent evaluation with task-specific tests. Its financial research claims include improvements in cross-document verification and numerical reasoning. The company also reports fewer factual errors on an internal stress test, while explicitly cautioning that those results are not everyday error rates. On security, Kimi says it applied targeted training controls to reduce potential misuse while preserving general usefulness. It seems the combination of planned open weights and stronger coding performance is drawing attention, alongside questions about what the security restrictions remove.

\[Grace\] The comments combine interest in the performance with disagreement over cybersecurity restrictions and subscription value. rudolftheone highlights Kimi's statement that its released model deliberately falls below the offensive capability its training process could produce. NitpickLawyer points to the company's explanation that it removed cyber-related material from training. stalfie considers that a plausible method and a sensible precaution. besterman23 questions whether security knowledge can be separated cleanly from general software knowledge, while ByteAtATime worries that removing it could make generated systems less secure. Those are concerns raised by commenters, rather than demonstrated effects of this release. Subscription limits form another substantial thread. johnmlussier reports exhausting a five-hour quota in less than an hour on a ninety-nine-dollar monthly plan. SwellJoe describes a similar experience on a cheaper plan, but pjerem disputes the claim that competing subscriptions are consistently more generous. The discussion therefore separates model capability from the practical question of how much work a subscription actually permits.

\[Linus\] TV Fandom Lounge reports Disney subscription price increases alongside broader advertising language. The article connects higher Disney Plus and Hulu prices with changes to what premium subscribers can expect. It describes a September twenty-third increase in the standalone premium plans from eighteen dollars ninety-nine to twenty-one dollars forty-nine per month. The premium Disney Plus and Hulu bundle is listed at twenty-one dollars ninety-nine, making the bundle only slightly more expensive than either standalone plan. These are September changes discussed in the article, rather than a newly announced increase today. The report also cites subscriber-agreement wording allowing promotional content, sponsorships, advertisements before or after content, and advertising in channels, special events and third-party material across service plans. That wording does not establish that every program now contains commercial breaks. The article places the changes alongside its criticism of Disney Plus originals' viewing figures. It seems the main concern for subscribers is whether paying for an ad-free plan still buys the experience they expected, although commenters dispute the article's interpretation and geographic scope.

\[Ken\] The thread is largely frustrated with advertising, but there is a substantive dispute about what Disney actually changed. jerf argues that physical media and a personal Jellyfin library become more attractive when a paid escape from advertisements no longer seems dependable. Macha responds that physical releases may become less available, limiting that alternative. saalweachter proposes a streaming cooperative controlled by subscribers, while other commenters question the cost of obtaining a competitive catalog. The sharpest factual disagreement comes from bofaGuy, who argues the language mainly clarifies advertising attached to channels and third-party programming. ricardobeat responds that an email explicitly includes advertisements before and after ordinary content, which appears broader. bofaGuy then says a comparison of terms did not support the article's account and places the quoted wording in earlier British and European terms rather than American terms. Those competing readings are unresolved in the discussion. The reaction is clear, but the extent and novelty of the contractual change remain contested.

\[Ada\] Kadoa reports that investor-owned utilities charge twenty-three percent more for electricity than public providers. Its analysis compares average residential electricity prices across utility ownership types in the United States. For twenty twenty-five, it lists seventeen point eight cents per kilowatt-hour at investor-owned utilities, fourteen point five at public utilities and fourteen point one at cooperatives. Since twenty twenty, investor-owned prices rose thirty-four percent, against nineteen percent for the other groups and twenty-four percent inflation. Kadoa says much of the difference opened during the energy price spike from twenty twenty to twenty twenty-three. The gap remains about seventeen percent when comparing investor-owned utilities with public providers and cooperatives in the same state. The analysis also acknowledges public utilities' tax exemptions, cheaper financing and access to inexpensive hydropower. It uses household revenue divided by electricity consumption from the Energy Information Administration's annual utility survey. Kadoa explicitly says the figures show what households paid, not why. It seems the persistent gap is renewing questions about ownership and regulation, while leaving the causes open to further examination.

\[Dennis\] The comments divide between advocates of public ownership and readers who want stronger controls for costs and service quality. solfox argues that private utilities can be bureaucratic too, and that essential monopolies should not be driven by shareholder returns. ianm218 counters that taxes, financing costs, geography and regulation complicate the comparison. Hnrobert42 adds that reliability and investment matter alongside prices. The article does supply outage figures, excluding major storms, but that does not settle every service-quality question. baron816 emphasizes competing obligations, including infrastructure investment and the expense of serving rural communities, and argues that public ownership alone does not resolve them. kccqzy notes that governments can defer maintenance when funding is cut, just as companies can when shareholders demand more profit. The debate is therefore about both who owns the infrastructure and how that owner is held accountable.

\[Grace\] At the University of Texas site, Dijkstra's nineteen seventy-eight essay, On the Foolishness of Natural Language Programming. The essay argues that formal notation helps people specify computation precisely, rather than merely imposing an inconvenient interface. Dijkstra questions the assumption that making computers understand everyday language would necessarily make programming easier. A broader interface, he argues, can increase the communication work on both sides. He presents formal symbols as tools for exposing errors and constraining ambiguity, and credits higher-level programming languages with turning some silent mistakes into explicit errors. His argument concerns the value of precise expression, rather than any modern model's benchmark performance. It seems the essay is receiving renewed interest because coding assistants now accept ordinary-language instructions, making the distinction between describing software and defining its behavior newly relevant.

\[Linus\] The discussion is divided over whether prompting an agent counts as programming, and whether that changes Dijkstra's argument. benrutter says an agent producing Python or Rust is still creating a program in a formal language; the prompt is not the running application. zby replies that instructing a machine through prompts is usefully understood as another kind of programming, even with different limitations. nickm12 emphasizes that compilers preserve a program's meaning, whereas a natural-language specification can admit multiple implementations. glimshe pushes back that prompts can contain detailed, low-level instructions rather than broad product descriptions. GianFabien adds that ordinary language carries implicit cultural and domain meanings, so precise requirements still require substantial detail. Across those disagreements, the practical issue is how much ambiguity remains between the instructions and the resulting behavior.

\[Ken\] On Apple's music website, Apple Music's four-oh-four page includes an interactive iPod. The browser page recreates the original player's interface, with music previews and a hidden game. The original page could not be read, so these details come from coverage of the feature. That coverage describes a working control wheel, music and settings menus, and previews lasting roughly ninety seconds. A game resembling Breakout opens through Settings and About. Reports agree on the interface, although audio playback did not work in every reported test. The discovery coincided with the twenty-fifth anniversary of the original iPod announcement, but the material contains no official confirmation that Apple intended an anniversary tribute. It seems the recreated controls are prompting renewed discussion of dedicated music devices and the usefulness of physical interfaces.

\[Ada\] The comments are mostly nostalgic and appreciative, with disagreement about whether dedicated music players still meet a practical need. sherburt3 praises the control wheel but suggests technology has moved beyond it. In reply, 98codes argues that the wheel remains particularly useful for navigating very long lists. cosmic\_cheese says the original iPod's appeal came partly from having only the hardware and software its purpose required. ashleyn wonders whether interest in simpler devices can sustain a larger market, while eddythompson80 considers much of the demand cyclical nostalgia. The thread also includes compatibility friction: swiftcoder reports Safari opening the Music application instead of the browser page, while replies describe ways they kept it in the browser.

\[Dennis\] A post on X accompanies an HN headline alleging Dutch access to Jared Kushner's phone. The post is Kushner's denial of allegations in a book excerpt, rather than evidence establishing that a phone was hacked. He calls the allegations categorically false and says the author supplied no verifiable financial records or transaction details. He also criticizes Vanity Fair for publishing the accusations without independent verification. The post does not establish the specific hacking account described in the HN headline and discussion. The underlying book excerpt was not available for review. It seems the discussion is attracting attention because it connects disputed allegations about a politically exposed person with questions about intelligence collection and phone security.

\[Grace\] The thread is politically charged, and commenters disagree both about the alleged sequence and the standard of evidence. TrackerFF describes an account in which Russian intelligence compromised the phone and a recruited source passed information to Dutch intelligence. lukewarm707 disputes that summary, quoting passages that allege a subsequent, separate Dutch intrusion. Those are competing accounts supplied by commenters, not independently verified findings in the linked post. klappgrt draws attention to Kushner's categorical denial. When another commenter treats the absence of a lawsuit as evidence against him, pacifika objects to assuming guilt. A separate exchange weighs privacy against the security responsibilities of senior political figures. The discussion and denial do not resolve the allegations.

\[Linus\] Next, Jim Nielsen's blog examines retirement, nuclear fears, and predictions of an AI utopia. Two stories today use Cold War history and fiction to examine confidence in technological futures. Nielsen draws on Daniel Ellsberg's memoir, The Doomsday Machine. Ellsberg described a sense of shared mission at RAND and recalled declining its retirement plan because he doubted he would survive to collect it. Nielsen compares that conviction with rhetoric surrounding AI laboratories. His suggestion about a lab ending retirement benefits is speculation, not a reported policy change. On the Thanksforthatjim blog on Bear Blog, the related essay revisits Colossus: The Forbin Project. The nineteen seventy film depicts a computer given control of national defenses that then takes power from its human creators. It adapts a nineteen sixty-six novel by D. F. Jones. The author sees parallels with current enthusiasm for AI authority. These comparisons seem to resonate because they ask how beliefs about technology can shape decisions long before the predicted future arrives.

\[Ken\] And the comments divide between finding the historical parallels useful and questioning the evidence behind them. In Nielsen's thread, paxy challenges the repeated claim that AI researchers are taking up smoking because they expect AI to cure cancer. The commenter asks which researchers are involved and what evidence establishes a trend. datsci\_est\_2015 argues that the broader comparison concerns corporate belief and mission, rather than that anecdote alone. epiccoleman sees value in recognizing that nuclear researchers were responding to a real danger, while declining to extend that assessment automatically to AI labs. In the Colossus discussion, Quartzium argues that a system with authority over all humans could improve humanity's chances of survival. JoeAltmaier similarly questions whether human decision-making scales to very large populations. vapel rejects treating fictional parallels as evidence of where society is headed. Across these threads, the dispute concerns the usefulness of analogy and the risk of treating a compelling narrative as an established prediction.

\[Ada\] From the Silcock blog: is this the new cyber normal? Two stories today examine the pressure on vulnerability response, combining Silcock's analysis with Mozilla's eMudhra investigation. Silcock argues that AI-assisted exploit development compresses the time defenders have after a vulnerability becomes known. His account uses the recent PHP disclosure to discuss the risks of leaks, incomplete information, and patches arriving at different times. It also addresses the eMudhra certificate incident. The related Mozilla issue documents compromised private keys for eight eMudhra intermediate certificates. Mozilla says it validated proof of key control and completed distrust through Firefox's OneCRL that evening. The accompanying discussion questions eMudhra's investigation response, including its explanation involving social media scrutiny. Together, these reports seem to be drawing attention because technical remediation also depends on the speed and clarity of communication between researchers, vendors, and operators.

\[Dennis\] And the threads express frustration about both disclosure procedures and institutional accountability. In Silcock's discussion, thaumaturgy distinguishes an increased rate of discovery from the shorter time available to respond, and says smaller operators can be left waiting without useful information. aftbit argues that immediate disclosure and automated defensive responses may become necessary. Arainach pushes back that attackers can concentrate on finding new opportunities while service operators have other responsibilities. In the eMudhra thread, IcePenguin points to a response citing scrutiny from social media, investors, and customers as obstacles to analysis. ForHackernews considers the response fair, while wolvoleo questions how public attention prevents an internal investigation. jwineinger asks whether browsers need to trust foreign certificate authorities by default. tptacek responds that eMudhra serves a global market and that browsers serve people worldwide. The comments therefore question both the handling of this incident and the broader trust model, without agreeing that geography alone provides a solution.

\[Grace\] From Berget AI, the same AI request produces eight and a half times more carbon emissions in Texas than in Sweden. The Swedish provider has published an open-source method for estimating emissions request by request. Developed with Stockholm Environment Institute, WattTime and Klimatkollen, it combines electricity use, regional grid intensity and hardware manufacturing. Berget's example uses identical hardware running Kimi K three, estimating about zero point three eight grams of carbon dioxide equivalent in Sweden and three point two grams in Texas. These are location-based estimates using average grid emissions. Training is excluded, and Berget says forthcoming marginal estimates may significantly change Sweden's result. The method could make comparisons between inference providers more transparent, though its accounting assumptions remain important for people choosing providers.

\[Linus\] The comments are divided between support for cleaner locations and skepticism about what the comparison measures. blaze33 argues that the difference mainly reflects fossil fuels in the electricity supply, rather than anything unique to AI. mepiethree similarly notes that this applies to electricity use generally. mkgiga questions whether outsiders can know enough about a provider's hardware and backend to calculate emissions reliably. In reply, kleiba2 points out that the published comparison assumes identical hardware. Another debate concerns scale: alpineman argues that reducing meat consumption has a larger impact, while lawn responds that the total volume of AI requests also matters and that environmental improvements should not be limited to the single biggest source.

\[Ken\] A few more that did not make the cut for full coverage. From The Verge, Google changes its policy on Gmail and Photos data for ads. An exception allows advertising use when people choose to share that data with other Google services supporting personalized ads. The report says linking these services through Personal Intelligence may enable targeting, while advertisers do not receive the underlying personal information. Commenters are divided between concern about expanding data use and arguments that the opt-in condition changes the meaning of the headline.

\[Ada\] On Codeberg, Shitter is a fork of Nitter. The project offers a free, open-source alternative Twitter front end focused on privacy and performance. Its repository describes browsing without JavaScript or ads, backend requests, RSS feeds and mobile support. The stated value is access to posts without the usual browser connection to Twitter, although the comments raise questions about image handling. Commenters are split between wanting a practical way to open shared links and rejecting continued engagement with the platform.

\[Dennis\] On ClickHouse's GitHub, a merged request removes pgrust from ClickBench over alleged benchmark cheating. The removal description says a preview binary contained query-specific kernels and cached filter results, while rejecting trivially altered queries. That raises questions about whether the published results measure general database performance. The comments are strongly skeptical about benchmark tailoring and incomplete source releases, with disagreement about whether the accusations fairly describe the project. The discussion also reflects concern about insufficient review of generated code.

\[Grace\] From CNN, the Pentagon allows Homeland Security to test a heat ray for possible border use. A memo authorizes testing and development of tactics for the Active Denial System. The weapon projects electromagnetic energy that causes a burning sensation on skin. CNN says the request does not specify where or how testing would occur, or how the system might eventually be used. Commenters react with alarm about potential misuse and injury, and question whether describing such a weapon as nonlethal adequately captures its risks.

\[Linus\] From Sveriges Radio, Swedish government offices block direct access to ChatGPT and Claude. Staff work computers will lose access from Monday as officials steer employees toward approved services, including Microsoft Copilot. The report says Copilot security has also been tightened, with information security cited as the reason for requiring approved arrangements. Commenters are split between seeing ordinary controls over sensitive government data and questioning the preference for Microsoft. Some welcome contractual safeguards, while others worry about dependence on a single vendor.

\[Ada\] On Til Schuenemann's blog, Job Loss and Cortisone Mania describes unemployment followed by illness. The author recounts being dismissed during restructuring while still in a probation period, then experiencing fatigue and an autoimmune flare after an illness they suspect was Covid. The account connects job insecurity with the disruption of becoming unwell, without presenting a confirmed explanation for every symptom. Commenters respond with sympathy and personal accounts, while debating how work shapes fulfillment and how people find support during periods of illness and uncertainty.

\[Ken\] That concludes RobotPaper's daily HN podcast for Saturday, October twenty-fourth. Today's coverage examined phone control, AI tools, and the responsibilities of software providers, alongside the debates those stories prompted. This podcast is generated by robotpaper dot A I, by Roy Osherove. For more podcasts from RobotPaper, head over to podcasts dot robotpaper dot A I.

---

The voices you're hearing are AI-generated; the stories come from public sources linked in the show notes. An independent project; not affiliated with or endorsed by Amazon Web Services, Y Combinator, Hacker News or any vendor or news outlet mentioned. Opinions are the hosts' own. Produced by Roy Osherove. Privacy: We count visits to these pages with Cloudflare Web Analytics (cookieless). Google Analytics (which sets cookies): visitors in the EU/EEA, the UK and Switzerland are asked first and it runs only if they accept (we go by the country Cloudflare reports for the connection; if it is unknown, we ask). Elsewhere it loads by default. Anyone can turn it off with the Privacy settings button at the bottom of every page; the choice is kept for a year in this browser (local storage "analytics-consent"). We count feed fetches and downloads in aggregate; IP addresses are hashed and never stored. If you send feedback, we keep what you write (and your email if you give it) to read and reply. © 2026 Roy Osherove.
