Philippe Artières, historien : «Le caractère massif des dossiers Epstein nous fait croire à leur exhaustivité»
Le directeur de recherches au CNRS tente de qualifier les millions de documents révélés par la justice américaine sur les agissements du criminel sexuel. Interview :
That’s one hell of a cover…..
Mark Simonson reminisces about when he discovered type design. “The idea of coming up with an original alphabet design fired my imagination. And learning that it was possible to design type professionally was a revelation.” [marksimonson.com]
You must picture Sam Altman holding a knife to Tim Berner-Lee's throat. We are in a truly existential level of danger when it comes to the survival of the open web, across every front. This year is when it all comes to a head. www.anildash.com/2026/03/27/e...
Remembering Aretha Franklin on her birthday.
💜 🙏🏼 🎵 🎹🎤
No really, I am not kidding when I say that the data broker industry must be destroyed: www.npr.org/2026/03/25/n...
Look at us go! Now on YouTube!
Major credit to @rhodesben.bsky.social who spent 90 minutes answering @wired.com questions about geopolitics for a fascinating, extra-long episode of Tech Support this week:
By ditching encryption, Meta is making a decision to put people's lives in danger all over the world
A teenager already went to jail because of Facebook messages about abortion.
Human rights activists have been tortured and murdered due to their messages being obtained
Encryption is life or death
"He forgets to eat, he says, unless his wife puts food in front of him. His back hurts. His eyes are constantly tired. He drinks too much coffee. Some mornings he forgets what day it is or when he last went outside."
The Washington Post doing surveillance pricing was not on my 2026 bingo card, and if it were on anyone’s it would have been on mine
For years, we warned that the case against Julian Assange could be used against mainstream journalists, despite what many pundits said.
Well … now, it’s happening.
Watch @trevortimm.bsky.social expose how the Trump administration is citing the WikiLeaks case in court to target journalists’ rights.
Jayapal: To the survivors in the room, if you are willing, please stand. And if you are willing, please raise your hands. If you have still not been able to meet with this DOJ. Please note for the record that every single survivor has raised their hand.
Today we’re launching Open Climate Risk, a fully open option for U.S. building-level climate risk data. It’s unique because it allows you to see not only risk scores, starting with wildfire, but also the complete underlying dataset, methods, and codebase. carbonplan.org/research/cli...
Bending Spoons remains largely unknown, even as its portfolio of products has served more than a billion people.
Putin pioneered this. No opposition is legitimate. Regular, decent, “ordinary” people cannot possibly be against us because only we represent “the real people.”
Tell-tale sign of authoritarianism.
Officials in authoritarian regimes often make obviously false assertions to signal loyalty to the ruler and to determine who shares that loyalty
No he didn’t, Libby. His written remarks referred to Greenland as a “piece of ice” because that’s what it is. You’re the only one mixing anything up here.
“The Party told you to reject the evidence of your eyes and ears. It was their final, most essential command.”
Anthropic published their “soul document”
This is a continuation of “constitutional AI”. The constitution document is now a large document of prose that’s used in s number of training stages, even synth data generation as well as RL & SFT
(Strix confirmed fwiw)
www.anthropic.com/news/claude-...
If you tell an AI to convince someone of a true vs. false claim, does truth win? In our *new* working paper, we find...
‘LLMs can effectively convince people to believe conspiracies’
But telling the AI not to lie might help.
Details in thread
Canadian Prime Minister Mark Carney warned that world order is at a “rupture” point due to the U.S.’s longstanding vise-grip on the world and its swiftly expanding authoritarian nature under President Donald Trump.
Research shows that more diverse juries are less quick to convict, and that they deliberate longer and more carefully.
A landmark reform by Governor Phil Murphy on Sunday will make the jury pool in New Jersey a lot more representative of the state.
Why do so many apparently healthy people die or get severely ill from infections that would be considered low risk?
For one, we don't have any way of assessing a person's immune system function in the clinic
nature.com/articles/d41...
High RAM prices mean record-setting profits for Samsung and other memory makers arstechnica.com/gadgets/2026...
In 2024, I rose in Senate QP to ask why the Govt of Canada was still using X for official communications. Today, they are still there. What exactly will it take to break the spell?
Don’t work for a company you would not invest money in, because when you are working you are investing the most valuable thing you have: your time.
For more advice see my book Excellent Advice for Living amzn.to/3mQL4c4 #excellentadvice
A large comparison table showing AI model pricing and benchmark performance across many tasks. Column headers (left to right): Benchmark | Description | Gemini 3 Flash (Thinking) | Gemini 3 Pro (Thinking) | Gemini 2.5 Flash (Thinking) | Gemini 2.5 Pro (Thinking) | Claude Sonnet 4.5 (Thinking) | GPT-5.2 (Extra high) | Grok 4.1 Fast (Reasoning) Pricing rows (top): • Input price ($/M tokens): Gemini 3 Flash $0.50 | Gemini 3 Pro $2.00 | Gemini 2.5 Flash $0.30 | Gemini 2.5 Pro $1.25 | Claude Sonnet 4.5 $3.00 | GPT-5.2 $1.75 | Grok 4.1 $0.20 • Output price ($/M tokens): Gemini 3 Flash $3.00 | Gemini 3 Pro $12.00 | Gemini 2.5 Flash $2.50 | Gemini 2.5 Pro $10.00 | Claude Sonnet 4.5 $15.00 | GPT-5.2 $14.00 | Grok 4.1 $0.50 Selected benchmark rows (scores shown as percentages unless noted): • Humanity’s Last Exam: Gemini 3 Flash 33.7% / 43.5%, Gemini 3 Pro 37.5% / 45.8%, Gemini 2.5 Flash 11.0%, Gemini 2.5 Pro 21.6%, Claude 13.7%, GPT-5.2 34.5% / 45.5%, Grok 17.6% • ARC-AGI-2: Gemini 3 Flash 33.6%, Gemini 3 Pro 31.1%, Gemini 2.5 Flash 2.5%, Gemini 2.5 Pro 4.9%, Claude 13.6%, GPT-5.2 52.9% • GPQA Diamond: Gemini 3 Flash 90.4%, Gemini 3 Pro 91.9%, Gemini 2.5 Flash 82.8%, Gemini 2.5 Pro 86.4%, Claude 83.4%, GPT-5.2 92.4%, Grok 84.3% • AIME 2025: Gemini 3 Flash 95.2% / 99.7%, Gemini 3 Pro 95.0% / 100%, Gemini 2.5 Flash 72.0% / 75.7%, Gemini 2.5 Pro 88.0%, Claude 87.0% / 100%, GPT-5.2 100%, Grok 91.9% • MMMU-Pro: Gemini 3 Flash 81.2%, Gemini 3 Pro 81.0%, Gemini 2.5 Flash 66.7%, Gemini 2.5 Pro 68.0%, Claude 68.0%, GPT-5.2 79.5%, Grok 63.0% • ScreenSpot-Pro: Gemini 3 Flash 69.1%, Gemini 3 Pro 72.7%, Gemini 2.5 Flash 3.9%, Gemini 2.5 Pro 11.4%, Claude 36.2%, GPT-5.2 86.3% (with python) • CharXiv Reasoning: Gemini 3 Flash 80.3%, Gemini 3 Pro 81.4%, Gemini 2.5 Flash 63.7%, Gemini 2.5 Pro 69.6%, Claude 68.5%, GPT-5.2 82.1% • OmniDocBench 1.5 (lower is better): Gemini 3 Flash 0.121, Gemini 3 Pro 0.115, Gemini 2.5 Flash 0.154, Gemini 2.5 Pro 0.145, Claude 0.145, GPT-5.2 0.143 …
Gemini 3 Flash
Basically Gemini 2.6 Pro but with Flash pricing & speed
blog.google/technology/d...
