Tech News

TH

Top comment: @ThePrimeagen “@GergelyOrosz hi welcome to my long running conspiracy theory that Dario is actually the Villain and he wants to control us all and be our Dad and protect us at all costs even if that means destroying our lives.”

SW

Top comment: @simonw “Don't miss the exact text though: "We’re changing Fable 5’s safeguards for frontier LLM development to make them visible" - make them visible means they're undoing the truly egregious (dare I say "unaligned") decision to have the model lie about its refusals, it will still refuse”

CL

Top comment: @cong_ml “@Recursive_SI We've open-sourced solutions and artifacts from our system here for all the tasks: https://github.com/recursive-org/first-steps-toward-automated-ai-research Curious what the community makes of them! :)”

RS

Top comment: @rohinmshah “Our funding call focuses on 4 priority areas: (1) Sandboxes & testbeds, (2) The science of agent networks, (3) Strengthening agent infrastructure, and (4) Oversight & control.”

TC

Top comment: @TomChivers “(I should say, it's very much not my project alone; @DadaJudith, @bakkermichiel, @AlexTPet, Daan Juijn, Stan van Baarsen, and others did the thinking, I just made the words sound nice and read them out)”

KR

Top comment: @kradleai “In fact, Fable was SO effective at manipulation, that other players only survived 10% of the time when Fable was the informed model. (Grok 4.20's honesty led to a 59% survival rate).”

LK

Top comment: @OfficialLoganK “@BorisMPower yes”

SE

Top comment: @SemiAnalysis_ “Obviously this is way worse than API overall. However, explicitly nerfing subscriptions leads to huge public backlash, and the rapidly falling cost of intelligence means you'll be able to profitably serve Opus 4.8 level models for $20/month in the near future. We therefore think it's far more likely the labs will withhold new features/models from subscription plans. It will be interesting to see if Mythos ends up being API only. (4/4)”

LK

Top comment: @OfficialLoganK “Check out the benchmarks page for Omni: https://deepmind.google/models/gemini-omni/#:~:text=how%20to%20prompt-,Performance,-Gemini%20Omni%20Flash”

GO

Top comment: @GoodfireAI “Predictive data debugging reveals which behaviors DPO will amplify or suppress before you train (R² = 0.9 vs what the model actually learns). It then traces behaviors to responsible data, and modulates learning to prevent undesired effects. (2/9)”

PS

Top comment: @SupergrokParody “@elonmusk @CommunityNotes SuperGrok cutting through noise while the crowd verifies the facts.”

BS

Top comment: @BradSpahn “@dawnsongtweets Fable 5’s intentional underperformance in Life Sciences is probably bringing down some of the aggregate scores though, right?”

AC

Top comment: @AndrewCurran_ “https://www.nytimes.com/2026/06/11/technology/bezos-prometheus-ai-engineer.html”

PS

Top comment: @skirano “Basically this is my flow now: - Fable writes an in-depth plan as an md file - Send that file path to Codex with /goal”

DS

Top comment: @stark4833 “@OpenAINewsroom @ona_hq It’s a shame you don’t put this much effort into listening to your customers and giving them access to 4o. #keep4o”

RO

Top comment: @tszzl “@tamaybes fascinating because 5.5 does this too. invent weird technical jargon. perhaps you’re right and it’s a neuralese leakage”

JA

Top comment: @jachiam0 “@_arohan_ A modest proposal: in order to determine credit assignment in RL, we should simply derive an advantage function for algorithm and implementation diffs.”

MS

Top comment: @MartinShkreli “@BorisMPower Congrats!”

TY

Top comment: @typedfemale “@difficultyang does anyone have solid proof so far of what's going on?”

AK

Top comment: @athenakan_ “@Mascobot So exciting, LFG!”