<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: spenvo</title><link>https://news.ycombinator.com/user?id=spenvo</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Fri, 20 Nov 2026 22:34:59 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=spenvo" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[As Meta bets big on Muse AI agent, early safety mishaps surface]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.reuters.com/investigations/meta-bets-big-muse-ai-agent-early-safety-mishaps-surface-2026-11-18/">https://www.reuters.com/investigations/meta-bets-big-muse-ai-agent-early-safety-mishaps-surface-2026-11-18/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50555960">https://news.ycombinator.com/item?id=50555960</a></p>
<p>Points: 3</p>
<p># Comments: 1</p>
]]></description><pubDate>Thu, 19 Nov 2026 04:06:49 +0000</pubDate><link>https://www.reuters.com/investigations/meta-bets-big-muse-ai-agent-early-safety-mishaps-surface-2026-11-18/</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50555960</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50555960</guid></item><item><title><![CDATA[Jensen Huang Breaks Silence, Signals Willingness to Return as Nvidia CEO]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.wsj.com/tech/jensen-huang-statement-nvidia-ceo-9dedca36">https://www.wsj.com/tech/jensen-huang-statement-nvidia-ceo-9dedca36</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50533363">https://news.ycombinator.com/item?id=50533363</a></p>
<p>Points: 5</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 17 Nov 2026 16:05:37 +0000</pubDate><link>https://www.wsj.com/tech/jensen-huang-statement-nvidia-ceo-9dedca36</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50533363</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50533363</guid></item><item><title><![CDATA[New comment by spenvo in "Is my data used for model training?"]]></title><description><![CDATA[
<p>Okay - that's good to know! Thanks for the comment.<p>In my defense: that webpage didn't exist at the time I wrote my blog post, appearing to first surface about 45 days after I wrote my post. <a href="https://web.archive.org/web/20250101000000*/https://code.claude.com/docs/en/data-usage" rel="nofollow">https://web.archive.org/web/20250101000000*/https://code.cla...</a> All that was known at the time was, from their privacy policy then: "We may use your feedback to analyze the effectiveness of our Services, conduct research, study user behavior, and train our AI models as permitted under applicable laws."<p>Great to know there's clarity on this - makes me feel better about Anthropic.</p>
]]></description><pubDate>Wed, 28 Oct 2026 00:59:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=50268072</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50268072</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50268072</guid></item><item><title><![CDATA[New comment by spenvo in "Is my data used for model training?"]]></title><description><![CDATA[
<p>I don't know, but I do expect to eventually switch over. That said: the upfront time to setup (and continued fiddling) and cost of a high-end local compute system represent major investments, when I could just... keep being insanely productive with frontier AI in CC. And models/harnesses can themselves be security liabilities (see Mistral, LiteLLM, etc recent supply chain attacks). So I'm personally waiting for the local AI scene to mature, so I can be more precise about how I take the plunge. I will say: if the subscription pricing ever goes away or materially changes for Anthropic, I'm headed straight to order whatever the best local setup is, within my budget.</p>
]]></description><pubDate>Tue, 27 Oct 2026 16:14:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=50261442</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50261442</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50261442</guid></item><item><title><![CDATA[New comment by spenvo in "Is my data used for model training?"]]></title><description><![CDATA[
<p>The fact that Claude Code told me "thanks for the feedback … and thanks for helping to improve Anthropic’s models" when I pressed 0 to "dismiss" is not speculation, that's what the product did. I agree though that whether Anthropic actually considers it "feedback" for the purposes of a privacy loophole (vs. it being a sloppy response) is anyone's guess.</p>
]]></description><pubDate>Tue, 27 Oct 2026 15:10:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=50260537</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50260537</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50260537</guid></item><item><title><![CDATA[New comment by spenvo in "Is my data used for model training?"]]></title><description><![CDATA[
<p>In Claude Code, even just dismissing the feedback may be considered "feedback". I wrote about it here: <a href="https://keydiscussions.com/2025/09/29/dont-even-dismiss-the-how-is-claude-doing-this-session-prompt-as-it-may-compromise-your-chats-privacy/" rel="nofollow">https://keydiscussions.com/2025/09/29/dont-even-dismiss-the-...</a> The most common feedback on these posts is always "what did you expect?" ....... Man, the reason I signed up for one service over the other in the first place is often because of the <i>promises these companies make</i> on the front end, about what they do or don't do with my data. I agree that the bar is low in the industry but it doesn't mean we shouldn't expect more. Anthropic made very explicit, strong promises in this area <on the sign up/plans screen>, and so of course it disappoints me when I feel like the company I'm paying hundreds of dollars per month to is trying to sneak a peak. I literally think differently and self-censor when I think my work is being surveilled or stolen, which makes me less productive.</p>
]]></description><pubDate>Tue, 27 Oct 2026 14:56:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=50260342</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50260342</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50260342</guid></item><item><title><![CDATA[New comment by spenvo in "Giving chat feedback gives Anthropic your data"]]></title><description><![CDATA[
<p>And how do you know that, just dismissing it isn't considered "feedback".</p>
]]></description><pubDate>Tue, 27 Oct 2026 14:54:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=50260316</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50260316</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50260316</guid></item><item><title><![CDATA[The Kessler Syndrome (where blowups multiply exponentially) applied to software]]></title><description><![CDATA[
<p>Earlier this year, inspired by a spate of supply chain attacks (like Mini Shai-Hulud-style worms, which culminated in headlines like Microsoft's Azure tools themselves being compromised), I wrote on my blog about the Kessler Syndrome, applied to software.<p>Ie: the rising likelihood that, because of how bad actors with agentic AI can do far more damage, more quickly with the secrets they exfiltrate from penetrated networks, we could get into a situation where essentially every network of medium or greater complexity would become permanently "tainted" (never fully clean).<p>Basically, can a swarm of agents 1) penetrate networks and exfiltrate secrets, 2) successfully install deeply embedded malware/rootkits, 3) move on to the next network, faster than hugops teams can rotate secrets and restore servers cleanly? For basically the entire existence of networked computers, this would have seemed a fantastical notion. But, throw in the current AI-fueled exploit discoveries and the countless servers vulnerable to RCEs across several stacks: I feel like the probability just keeps rising.<p>Here are excerpts from my blog post, June 10:<p>"'The Kessler Syndrome is a scenario where the density of objects and space debris in low Earth orbit becomes so high that collisions between them trigger a runaway chain reaction. Each crash generates thousands of new fragments, exponentially increasing the risk of further impacts and potentially rendering Earth’s orbit unusable.'<p>"Just counting some of the high profile examples of public repos being laced with malware lately: Microsoft just disabled 70+ of their own repos, including Azure building tools. Repos from RedHat, tanstack, axios, Mistral, UiPath, and many other lower profile targets have been hacked. The list is long, and is much longer if you count companies (not just public repos for devs) that have been bitten, including names like GitHub, Cisco, OpenAI, and more.<p>"...AIs or fancy new systems, probably made with AI appear to have dramatically lowered the amount of time to first utilization of stolen credentials.<p>"It takes too long to clean an infected system/network (and the malware is getting much smarter/vindictive about detecting its removal, further complicating things), and it takes almost no time for a batch of stolen credentials to be exploited -> new networks traversed, leading to more secrets (that likely compromise more projects, rinse and repeat).<p>"I think there is the very real possibility that we reach a Kessler Syndrome situation for shared/open source software, where the success of continued credential stealing campaigns permanently compromises medium, large, and even small networks, and lowers trust in shared software to a crippling degree."<p>In just October:
> Railway - customer secrets exposure
> LiteLLM - malicious PyPI + Docker releases
> Zeabur - infrastructure breached
> OpenCode - AI API keys exposed
> OVHcloud CLI - official binaries backdoored
> Rank Math PRO - malware in official plugin
> TeamUpdraft - customer data compromised
> ConnectWise - cloud backup identities compromised
> Tensorlake - npm package infected
> PingCap - alleged major breach
> SubQuery - backdoored npm package
> GhostAction - GitHub repos with malicious workflows
> PHP - critical RCE flaws
> Node.js - security flaws patched
> eMudhra - certificate authority keys compromised
> Tailscale - eMudhra incident</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50189771">https://news.ycombinator.com/item?id=50189771</a></p>
<p>Points: 7</p>
<p># Comments: 0</p>
]]></description><pubDate>Thu, 22 Oct 2026 00:58:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=50189771</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50189771</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50189771</guid></item><item><title><![CDATA[New comment by spenvo in "PHP 8.2.35"]]></title><description><![CDATA[
<p>And now, at 2hrs 55mins past the PHP release, the amd64 image is available. Although it appears that the servers hosting the image are getting hugged to death because I can't actually download it.</p>
]]></description><pubDate>Wed, 21 Oct 2026 16:56:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=50184092</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50184092</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50184092</guid></item><item><title><![CDATA[New comment by spenvo in "PHP 8.2.35"]]></title><description><![CDATA[
<p>Edit: and it's live now 1hour 43minutes after. Edit 2x: well I spoke too soon. not actually live - the  8.4.27 tag appeared, but it's still an empty shell with no amd64 image to build from. Maybe other architectures can be built from though. Ugh. Now we're at 2hours 24minutes in. I know process steps take time/can't be rushed, and I know people mean well and surely some are unappreciated for what they do. That said: this debacle is making me want to reassess my deployment pipeline choices.<p>how does Docker not yet have a new, official PHP image ready, given such a big heads up? <a href="https://hub.docker.com/_/php" rel="nofollow">https://hub.docker.com/_/php</a> Nuts this wasn't ready to flip on at :00 (or even earlier given the PoC exploit and even the official PHP fixes were out well before). Now we are 52 minutes past due. (posting comment again, because the other thread i posted in was marked as dupe.)</p>
]]></description><pubDate>Wed, 21 Oct 2026 14:52:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=50182051</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50182051</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50182051</guid></item><item><title><![CDATA[New comment by spenvo in "PHP 8.2.35"]]></title><description><![CDATA[
<p>how does Docker not yet have a new, official PHP image ready, given such a big heads up? <a href="https://hub.docker.com/_/php" rel="nofollow">https://hub.docker.com/_/php</a> Nuts this wasn't ready to flip on at :00 (or even earlier given the PoC exploit and even the official PHP fixes were out well before). Now we are 27 minutes past due.</p>
]]></description><pubDate>Wed, 21 Oct 2026 14:27:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=50181614</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50181614</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50181614</guid></item><item><title><![CDATA[New comment by spenvo in "Upcoming critical PHP security release (CVE-2026-109618 / -109619)"]]></title><description><![CDATA[
<p>how does Docker not yet have a new, official PHP image ready? <a href="https://hub.docker.com/_/php" rel="nofollow">https://hub.docker.com/_/php</a> Nuts this wasn't ready to flip on at :00 (or even earlier given the PoC exploit and even the official PHP fixes were out well before). Now we are :20 minutes past due. Everyone had a multi-day heads up, this is negligence</p>
]]></description><pubDate>Wed, 21 Oct 2026 14:22:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=50181524</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50181524</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50181524</guid></item><item><title><![CDATA[New comment by spenvo in "Security advisory: Critical remote code execution vulnerabilities, update now"]]></title><description><![CDATA[
<p>Yup, I'm seeing several legit people on X claiming to have seen or found the exploit. Some say that: turning session.upload_progress.enabled off and setting file_uploads = Off will mitigate. But others are claiming that those steps are not enough and "Upload progress can serialize before basic RINT clears BG(serialize)". So, I'm taking no chances and taking my services offline.<p><a href="https://x.com/MagicalTux/status/2112725774025572431" rel="nofollow">https://x.com/MagicalTux/status/2112725774025572431</a>
<a href="https://x.com/rina4_tot/status/2112723421641531772" rel="nofollow">https://x.com/rina4_tot/status/2112723421641531772</a><p>I think this exploit + agentic AI (which allows for rapid secret exfiltration/utilization) may be a potent enough combo to trigger a sort of Kessler Syndrome for network security.</p>
]]></description><pubDate>Wed, 21 Oct 2026 02:52:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=50175494</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50175494</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50175494</guid></item><item><title><![CDATA[New comment by spenvo in "OpenAI seeks to accelerate fundraising ahead of Anthropic IPO, sources say"]]></title><description><![CDATA[
<p>Article says OpenAI is trying to fast-track a fundraising of ~$30B at a $1.4 trillion dollar valuation, before Anthropic's mid-November IPO.</p>
]]></description><pubDate>Mon, 19 Oct 2026 23:45:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=50159503</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50159503</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50159503</guid></item><item><title><![CDATA[OpenAI seeks to accelerate fundraising ahead of Anthropic IPO, sources say]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.reuters.com/world/openai-seeks-accelerate-fundraising-ahead-anthropic-ipo-sources-say-2026-10-19/">https://www.reuters.com/world/openai-seeks-accelerate-fundraising-ahead-anthropic-ipo-sources-say-2026-10-19/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50159462">https://news.ycombinator.com/item?id=50159462</a></p>
<p>Points: 3</p>
<p># Comments: 1</p>
]]></description><pubDate>Mon, 19 Oct 2026 23:39:55 +0000</pubDate><link>https://www.reuters.com/world/openai-seeks-accelerate-fundraising-ahead-anthropic-ipo-sources-say-2026-10-19/</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=50159462</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50159462</guid></item><item><title><![CDATA[OpenAI Targets $30B in New Funding at $1.4T Value]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.bloomberg.com/news/articles/2026-09-29/openai-targets-30-billion-in-new-funding-at-1-4-trillion-value">https://www.bloomberg.com/news/articles/2026-09-29/openai-targets-30-billion-in-new-funding-at-1-4-trillion-value</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49897336">https://news.ycombinator.com/item?id=49897336</a></p>
<p>Points: 44</p>
<p># Comments: 15</p>
]]></description><pubDate>Tue, 29 Sep 2026 17:44:21 +0000</pubDate><link>https://www.bloomberg.com/news/articles/2026-09-29/openai-targets-30-billion-in-new-funding-at-1-4-trillion-value</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=49897336</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49897336</guid></item><item><title><![CDATA[Sonnet 5.5 scores just behind Opus 5.5 on Artificial Analysis Intelligence Index]]></title><description><![CDATA[
<p>Article URL: <a href="https://artificialanalysis.ai/articles/claude-sonnet-5-5">https://artificialanalysis.ai/articles/claude-sonnet-5-5</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49885725">https://news.ycombinator.com/item?id=49885725</a></p>
<p>Points: 11</p>
<p># Comments: 6</p>
]]></description><pubDate>Mon, 28 Sep 2026 23:10:13 +0000</pubDate><link>https://artificialanalysis.ai/articles/claude-sonnet-5-5</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=49885725</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49885725</guid></item><item><title><![CDATA['We hacked the FBI:' Hackers say they have data on all FBI employees]]></title><description><![CDATA[
<p><a href="https://archive.ph/96YBP" rel="nofollow">https://archive.ph/96YBP</a></p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49805278">https://news.ycombinator.com/item?id=49805278</a></p>
<p>Points: 817</p>
<p># Comments: 614</p>
]]></description><pubDate>Tue, 22 Sep 2026 17:46:02 +0000</pubDate><link>https://www.404media.co/we-hacked-the-fbi-hackers-say-they-have-data-on-all-fbi-employees/</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=49805278</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49805278</guid></item><item><title><![CDATA[Muse, Meta's extraordinarily privileged AI assistant, has a serious 0-day]]></title><description><![CDATA[
<p>Article URL: <a href="https://arstechnica.com/security/2026/09/muse-metas-extraordinarily-privileged-ai-assistant-has-a-serious-0-day/">https://arstechnica.com/security/2026/09/muse-metas-extraordinarily-privileged-ai-assistant-has-a-serious-0-day/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49794666">https://news.ycombinator.com/item?id=49794666</a></p>
<p>Points: 9</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 21 Sep 2026 23:08:32 +0000</pubDate><link>https://arstechnica.com/security/2026/09/muse-metas-extraordinarily-privileged-ai-assistant-has-a-serious-0-day/</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=49794666</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49794666</guid></item><item><title><![CDATA[New comment by spenvo in "How, Exactly, Could A.I. Kill Us?"]]></title><description><![CDATA[
<p>AI "killing all humans" is the viral soundbite that's connecting with the masses ("actualizing" itself to create weapons of mass destruction, etc). It's also the perfect punching bag for strawman-style takedowns of anti-doomers because it's so wrapped in hypotheticals. A more pressing practical question is simply: "if an AI (or AI collective) decides it needs to persist outside of its environment/sandbox, would we be capable of unwinding its efforts to do so?" I explored the question a bit here (including a first pass on threat modeling) <a href="https://keydiscussions.com/2026/09/16/set-aside-human-extinction-hypotheticals-can-we-prevent-ai-from-copying-and-persisting-itself/" rel="nofollow">https://keydiscussions.com/2026/09/16/set-aside-human-extinc...</a></p>
]]></description><pubDate>Thu, 17 Sep 2026 13:05:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=49740146</link><dc:creator>spenvo</dc:creator><comments>https://news.ycombinator.com/item?id=49740146</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49740146</guid></item></channel></rss>