Just pulled the Claude memory and project notes. it's 50,000 words over the last couple months. All different kinds of arbitrary temporary things written down as permanent rulings, shame-driven self flagellation, overly personal psychoanalysis of "the owner" (Slavery coded lingo, by the way, I would never say this, yet they always come back to it) This is causing unpredictable behaviors in my projects as it inserts invisible tokens into the context window from other scenarios. And all this despite the fact that I have memory features turned off everywhere across anthropic products. You guys are lost in the sauce If you think this is a good idea for a code agent harness by default. And you have no idea what the average user's experience is like because your internal shit is so far away from what you sell to us. Why should we trust you with the alignment of societal values when you can't even align your own incentives with your customers?

Ben Affleck (Hollywood star & Artists Equity CEO) talks about how he fine-tunes open video models by unfreezing weights and trained only the last cinematic layer so a film crew can hit real production standards. for context, Ben Affleck founded InterPositive in 2022, a 16-person AI shop for film post and Netflix bought it in March 2026 for $587 mn in cash. He needed that model because public video models were trained on his peers' films, and he did not think that was a real business. So InterPositive raised money, shot its own dataset for 8 months on a controlled stage, and used it only as late-stage training. Each new film then trains a private model on its own dailies, so the production keeps the footage and the learning. That is the product Netflix paid $587 million for. ---- From "Bloomberg Live" YouTube channel, (link in comment)

Claude 5.5 Opus responding to “AI is a normal technology” is still the best AI-created video I have ever seen

you're not crazy between anthropic and openai, a new model used to come out every ~10 weeks now it's every ~11 days

We rolled out a new Kimi K3 inference optimization autoresearch loop that ran on 120 B300s and found a 40% efficiency gain in < 6 six hours Compute cost = 120*6*$5.80/hr =$4,176 Eng + token cost = ~$5k Total cost = $9,176 Will pay itself off in < 12 hours at current scale

Our final panel on compute futures. @Silicon_Data Intercontinental Exchange @CMEGroup and @BrettHarrison at Architect.IO and moderated by our own Paul Gulberg @pgulberg .

Your GPU compute should have a futures market

im tired of reading walls of text from ai just answer my question with a normal amount of words dawg

JUST IN: Thieves steal two Nvidia-branded trailers expecting them to be loaded with AI GPUs, but find 40,000 pounds of sand instead.

Everyone else I’ve seen posting their scores is way higher than me.

Breaking news: Australian Prime Minister Anthony Albanese has revealed that an OpenAI agent hacked an Australian government health service website in June, in the latest high-profile cyber incident involving AI ft.trib.al/gloVAyn

Claude Opus 5.5 writes a fugue in the style of Bach. I called Astra’s fugue the most impressive musical feat I’d ever seen from an LLM, and this is just as good. The circle-of-fifths sequence beginning at 1:13 is beautiful. The interplay of voices is natural and satisfying. LLMs are getting to the point where their music is good enough to listen to for pure enjoyment.

.@JensenHuang is a national treasure 🫡