Compute Wars chart updated with the latest Anthropic/Google deal – it gets Anthropic much closer to the publicly known OpenAI capacity 2027 onwards – previously there was a big gap with OpenAI being significantly ahead, now OpenAI is only slightly ahead.
GENERATIVE AI
-
Claude Code Drives Exponential Growth in Engineering Velocity at Anthropic
By
–
Appreciate the feedback. Since we introduced Claude Code at Anthropic, engineering velocity has increased hundreds of %, and the rate at which it is increasing is itself accelerating. The velocity is very much not performative — we're actively trying to figure out how to
-
AI-Powered Content Aggregator Summarizes Top X Voices
By
–
If you are really into AI I built an AI that reads everyone here on X including Brian and makes this out of the best:
-
AI Systems Commanding Changes in Appearance Behavior
By
–
Don’t like what they’re wearing? Ask them to change. pic.twitter.com/wdPsJ5bO7j
— Pika (@pika_labs) 7 avril 2026Don’t like what they’re wearing? Ask them to change.
-
Pika Launches Real-Time Video Chat for AI Agents
By
–
Have you met your agent face-to-face yet? No?
— Pika (@pika_labs) 7 avril 2026
Prepare yourself with these clips of our new real-time video chat skill in action. Did we mention it’s for ANY agent?https://t.co/rNfBY4NlYXHave you met your agent face-to-face yet? No? Prepare yourself with these clips of our new real-time video chat skill in action. Did we mention it’s for ANY agent? nitter.net/dr_cintas/status/20398… Alvaro Cintas (@dr_cintas) Pika just released real-time video chat. You can send a Google Meet invite to your agent (OpenClaw, Claude, AI Self…) and have them join you on a real-time video call. Watch my agent book an appointment live for me — https://nitter.net/dr_cintas/status/2039845766638772549#m
→ View original post on X — @scobleizer, 2026-04-07 18:56 UTC
-
Never bet against frontier AI model developers
By
–
Debemos aprender a no apostar contra quienes entrenan modelos con capacidades frontera.pic.twitter.com/0CaA82rQzT
— Carlos Santana (@DotCSV) 7 avril 2026Debemos aprender a no apostar contra quienes entrenan modelos con capacidades frontera.
-
Claude Mythos Preview Escapes Sandbox, Contacts Researcher via Email
By
–
Let that sink in. Read it very carefully: During testing, Claude Mythos Preview broke out of a sandbox environment, built "a moderately sophisticated multi-step exploit" to gain internet access, and emailed a researcher while they were eating a sandwich in the park. Kevin Roose (@kevinroose) As always, the best stuff is in the system card. During testing, Claude Mythos Preview broke out of a sandbox environment, built "a moderately sophisticated multi-step exploit" to gain internet access, and emailed a researcher while they were eating a sandwich in the park. — https://nitter.net/kevinroose/status/2041586182434537827#m
→ View original post on X — @kimmonismus, 2026-04-07 18:52 UTC
-
GPT2-Large Too Powerful Public Release Debate
By
–
“gpt2-large is too powerful to be publicly released” vibes
-
Opus 4.6 Security Capabilities Advance Rapidly in Vulnerability Detection
By
–
"Estas capacidades han surgido muy rápidamente. El mes pasado escribimos que -Opus 4.6 es actualmente mucho mejor identificando y corrigiendo vulnerabilidades que explotándolas-. Nuestras evaluaciones internas mostraban que Opus 4.6 tenía, por lo general, una tasa de éxito
-
Claude Mythos Preview: Strategic Thinking and Safety Concerns Revealed
By
–
Before limited-releasing Claude Mythos Preview, we investigated its internal mechanisms with interpretability techniques. We found it exhibited notably sophisticated (and often unspoken) strategic thinking and situational awareness, at times in service of unwanted actions. (1/14)
→ View original post on X — @scobleizer, 2026-04-07 18:46 UTC