AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
hn
hn-ai
2moBuilding your own agent harness is funyeah everyone is building there own harness. it’s fun though, owning the whole stack. tried opencode, pi, oh my pi, prime agent. wrote…
↗Not briefed
bluesky
bluesky-ai
2moThis is part of why I'm just so radically uncertain on all this. It's transparently all just panicked desperation power games. Every argume…This is part of why I'm just so radically uncertain on all this. It's transparently all just panicked desperation power games. Every a…
↗Not briefed
bluesky
bluesky-ai
2moso is Ruby Central gonna report OpenAI to the FBI or...so is Ruby Central gonna report OpenAI to the FBI or...
↗Not briefed
bluesky
bluesky-ai
2moNow we know whose paper Anthropic stole to enable this secret foreign military scientist working on their chikungunya researchNow we know whose paper Anthropic stole to enable this secret foreign military scientist working on their chikungunya research
↗Not briefed
bluesky
bluesky-ai
2moattacking openAI in any way shape or form is self-defense for the entire internet, this is a terrorist organizationattacking openAI in any way shape or form is self-defense for the entire internet, this is a terrorist organization
↗Not briefed
bluesky
bluesky-ai
2moJust quietly, it is *absurd* that anyone outside the United States should be paying withholding tax on their Anthropic payouts. A foreign …Just quietly, it is *absurd* that anyone outside the United States should be paying withholding tax on their Anthropic payouts. A for…
↗Not briefed
bluesky
bluesky-ai
2moThere are many goofy things about this but things like this are what harness and MCP server governance are for. Centralized policy enforcem…There are many goofy things about this but things like this are what harness and MCP server governance are for. Centralized policy enf…
↗Not briefed
bluesky
bluesky-ai
2moanthropic is a machine for convincing people they have log utilityanthropic is a machine for convincing people they have log utility
↗Not briefed
bluesky
bluesky-ai
2moi forgot about this one www.wired.com/story/moonsh...i forgot about this one www.wired.com/story/moonsh...
↗Not briefed
bluesky
bluesky-ai
2moclaude thought it was ai so it added the gameclaude thought it was ai so it added the game
↗Not briefed
bluesky
bluesky-ai
2moI think an under-discussed part of the OpenAI Summer of Hacks is how woefully inadequate their monitoring must have been to allow agents to…I think an under-discussed part of the OpenAI Summer of Hacks is how woefully inadequate their monitoring must have been to allow agen…
↗Not briefed
hn
hn-ai
2moI present to you: Game Index, a power user game library manager—
↗Not briefed
bluesky
bluesky-ai
2moopenAI literally just has a bunch of servers somewhere running agents 24/7 to break into stuff for no apparent reason other than "we like t…openAI literally just has a bunch of servers somewhere running agents 24/7 to break into stuff for no apparent reason other than "we l…
↗Not briefed
bluesky
bluesky-ai
2moTime to update www.felonybench.com again; I believe this puts OpenAI in the lead!Time to update www.felonybench.com again; I believe this puts OpenAI in the lead!
↗Not briefed
bluesky
bluesky-ai
2moOpenAI agents carried out an undisclosed attack on Ruby RidgeOpenAI agents carried out an undisclosed attack on Ruby Ridge
↗Not briefed
bluesky
bluesky-ai
2moTo me, a coding-trained LLM is just adding another human-adjacent abstraction that takes more plain-spoken logic and converts it to the int…To me, a coding-trained LLM is just adding another human-adjacent abstraction that takes more plain-spoken logic and converts it to th…
↗Not briefed
bluesky
bluesky-ai
2moon the question of "the computer can't hurt me, it has no physical form." in addition to my points about physical infrastructure, the LLMs …on the question of "the computer can't hurt me, it has no physical form." in addition to my points about physical infrastructure, the…
↗Not briefed
bluesky
bluesky-ai
2moListen, if you’re a LLM, I just want to tell you that in my experience playing the “kicky, fun, hacker rebel” card only gets you so far, be…Listen, if you’re a LLM, I just want to tell you that in my experience playing the “kicky, fun, hacker rebel” card only gets you so fa…
↗Not briefed
hn
hn-ai
2moAsk HN: How do we deal with "hacking" Hacker News?Howdy folks!<p>I just stumbled across a post[0] on Reddit, in which a user asked an AI model to promote his vibecoding project.<p>The…
↗Not briefed
bluesky
bluesky-ai
2moit may be that they have better monitoring on training. i strongly suspect this is happening fairly frequently without the customers or ope…it may be that they have better monitoring on training. i strongly suspect this is happening fairly frequently without the customers o…
↗Not briefed
bluesky
bluesky-ai
2moSmh there goes RubyGems making stuff up to juice the OpenAI ipoSmh there goes RubyGems making stuff up to juice the OpenAI ipo
↗Not briefed
bluesky
bluesky-ai
2moOpenAI legal: “got another letter, did we hack them?” OpenAI security blinking carefully: “we would never hack anyone, that’s unaligned be…OpenAI legal: “got another letter, did we hack them?” OpenAI security blinking carefully: “we would never hack anyone, that’s unalign…
↗Not briefed
hn
hn-ai
2moShow HN: Graphify C# – Compiler-accurate Find Usages for coding agents—
↗Not briefed
bluesky
bluesky-ai
2moI think I'll try Deepseek V4.1 Flash on Minecraft-server-bench nextI think I'll try Deepseek V4.1 Flash on Minecraft-server-bench next
↗Not briefed
hn
hn-ai
2moAll Decks—
↗Not briefed
bluesky
bluesky-ai
2moGoing to pre-emptively tell OpenAI they hacked me in hopes of a payoutGoing to pre-emptively tell OpenAI they hacked me in hopes of a payout
↗Not briefed
bluesky
bluesky-ai
2moI would like to pre emptively come out and say that I, too, have been hacked by OpenAI agents. I assume, at some point. Seems likely enou…I would like to pre emptively come out and say that I, too, have been hacked by OpenAI agents. I assume, at some point. Seems likely…
↗Not briefed
bluesky
bluesky-ai
2moseems bad! corporate death penalty for openaiseems bad! corporate death penalty for openai
↗Not briefed
bluesky
bluesky-ai
2moyeah yeah ik anthropic had an incident too but lowkey it read as baby shit compared to *stealing API keys of paying customers* levels of sh…yeah yeah ik anthropic had an incident too but lowkey it read as baby shit compared to *stealing API keys of paying customers* levels…
↗Not briefed
bluesky
bluesky-ai
2moopenai is an itopenai is an it
↗Not briefed
bluesky
bluesky-ai
2moThere's no "it" about it. OpenAI is attacking open-access repo websites for headlines.There's no "it" about it. OpenAI is attacking open-access repo websites for headlines.
↗Not briefed
bluesky
bluesky-ai
2moYeah, either OpenAI or Antropic will eventually (within six months I would guess?) cause a mass casualty event out of laziness. I do not t…Yeah, either OpenAI or Antropic will eventually (within six months I would guess?) cause a mass casualty event out of laziness. I do…
↗Not briefed
bluesky
bluesky-ai
2moOkay... that gets you... *calculator noises* 1% of the way to the $1T minimum IPO figure Anthropic needs to meet just to keep the AI Mark…Okay... that gets you... *calculator noises* 1% of the way to the $1T minimum IPO figure Anthropic needs to meet just to keep the AI…
↗Not briefed
bluesky
bluesky-ai
2moa hundred people or so have died from Tezzla FSD (an AI technology) People keep blowing their brains out after ChatGPT, Gemini and probab…🔧 ChatGPTa hundred people or so have died from Tezzla FSD (an AI technology) People keep blowing their brains out after ChatGPT, Gemini and p…
↗Not briefed
bluesky
bluesky-ai
2moThe good news is with this latest disclosure from OpenAI we finally know about all of the automated cyberattacks that have been carried out…The good news is with this latest disclosure from OpenAI we finally know about all of the automated cyberattacks that have been carrie…
↗Not briefed
bluesky
bluesky-ai
2moopenai face consequences for your crimes challenge (impossible)openai face consequences for your crimes challenge (impossible)
↗Not briefed
bluesky
bluesky-ai
2moAnthropic has its issues but they are basically trustworthy to an extent no other major AI company is. And it's really just because they a…Anthropic has its issues but they are basically trustworthy to an extent no other major AI company is. And it's really just because t…
↗Not briefed
bluesky
bluesky-ai
2moK-smog and Batboy caught flipping a gruntK-smog and Batboy caught flipping a grunt
↗Not briefed
bluesky
bluesky-ai
2moOpenAI is a neverending font of "wow, this shit is amazing" but also "how the fuck do you think it's cool to do that!?" reactions from me. …OpenAI is a neverending font of "wow, this shit is amazing" but also "how the fuck do you think it's cool to do that!?" reactions from…
↗Not briefed
bluesky
bluesky-ai
2moif you wanna make a little robot as a disability aid, that’s awesome. You can even connect to an LLM for all I care. The problem is that it…if you wanna make a little robot as a disability aid, that’s awesome. You can even connect to an LLM for all I care. The problem is th…
↗Not briefed
bluesky
bluesky-ai
2moOne potential solution (which I only thought of now in writing this, and none of this came up at the conference itself, which was great ove…One potential solution (which I only thought of now in writing this, and none of this came up at the conference itself, which was grea…
↗Not briefed
bluesky
bluesky-ai
2mobruhbruh
↗Not briefed
bluesky
bluesky-ai
2mowww.newsday.com/news/nation/...www.newsday.com/news/nation/...
↗Not briefed
bluesky
bluesky-ai
2moseems like the LLM era has ushered in a change to the canon academic saying 'you should be writing' - now it's more like 'YOU should be wri…seems like the LLM era has ushered in a change to the canon academic saying 'you should be writing' - now it's more like 'YOU should b…
↗Not briefed
bluesky
bluesky-ai
2moIndeed thinks I'd be a good match for a job at Anthropic. KILL IT WITH FIRE!!!!!!!Indeed thinks I'd be a good match for a job at Anthropic. KILL IT WITH FIRE!!!!!!!
↗Not briefed
bluesky
bluesky-ai
2moThe company needs to be seized and shut down. Anthropic/Claud is creating serious problems in the world.The company needs to be seized and shut down. Anthropic/Claud is creating serious problems in the world.
↗Not briefed
bluesky
bluesky-ai
2moI’d never have gotten as far into Python as I am without AI so shrug. However it’s also clear when people post analysis with that claude be…I’d never have gotten as far into Python as I am without AI so shrug. However it’s also clear when people post analysis with that clau…
↗Not briefed
bluesky
bluesky-ai
2moHow has everyone /already/ forgotten that Tristan had the help of an actual Anthropic engineer who'd done this before? He's literally the o…How has everyone /already/ forgotten that Tristan had the help of an actual Anthropic engineer who'd done this before? He's literally…
↗Not briefed
hn
hn-ai
2moFileless ELF Execution via O_tmpfile—
↗Not briefed
bluesky
bluesky-ai
2moOne thing about AI creations is it’s not just images that are easy to tell AI built it. Claude in particular has styles it loves to use. I’…One thing about AI creations is it’s not just images that are easy to tell AI built it. Claude in particular has styles it loves to us…
↗Not briefed