link to @HeidyKhlaaf’s sharp analysis: Dr Heidy Khlaaf (هايدي خلاف) (@HeidyKhlaaf) As someone who has audited dozens of safety-critical systems, built static analysis tools, and used most formal verification and security tools, here are some red flags that should be a caution in taking these claims at face value: 1. There are no comparison benchmarks with 1/ — https://nitter.net/HeidyKhlaaf/status/2041591737563394442#m
RESEARCH
-
NousResearch CTO Audio Space Discussion on AI
By
–
By the way, I'll have NousResearch's CTO on an audio space today at 4 p.m. pacific. This helped me prepare, thanks!
-
Multiple Educational Resources Available Simultaneously for Learning
By
–
It's like having your kid go to ALL the best schools at one time.
-
Teaching AI by assembling world’s smartest experts
By
–
I taught it by giving it a football stadium of the world's smartest people to study. 🙂
-
Curating and Sharing Quality AI Content Daily
By
–
Your loss. I work hard to find the best AI posts to reshare.
-
Meta’s Muse Spark competes with frontier AI models on FrontierMath
By
–
Didn't expect Meta's model to outperform Opus 4.6 or GPT-5.4.
— Chubby♨️ (@kimmonismus) 8 avril 2026
But it still surprises me how well it performs and how close it is to other frontier models on toughest challenges. https://t.co/p7mKtDuj3f pic.twitter.com/TiQLRI5EHdDidn't expect Meta's model to outperform Opus 4.6 or GPT-5.4. But it still surprises me how well it performs and how close it is to other frontier models on toughest challenges. Epoch AI (@EpochAIResearch) We had pre-release access to Meta’s new Muse Spark model and evaluated it on FrontierMath. It scored 39% on Tiers 1-3 and 15% on Tier 4. This is competitive with several recent frontier models, though behind GPT-5.4. — https://nitter.net/EpochAIResearch/status/2041947954202988757#m
-
Open Models Discover FreeBSD Zero-Day Vulnerabilities
By
–
this is interesting.
— Gary Marcus (@GaryMarcus) 8 avril 2026
1. Did Anthropic forget to run a control?
2. Where does this leave us? https://t.co/frF8gNNrvOthis is interesting. 1. Did Anthropic forget to run a control? 2. Where does this leave us? Stanislav Fort (@stanislavfort) New post: We tested the Mythos showcase vulnerabilities with open models. They recovered similar scoped analysis! 8/8 models found the flagship FreeBSD zero-day, including a 3B model. Rankings reshuffle completely across tasks => the AI cybersecurity frontier is super jagged! — https://nitter.net/stanislavfort/status/2041922370206654879#m
-
Seven Years of Rapid Change in Artificial Intelligence
By
–
Incredible. Gary Marcus (@GaryMarcus) In AI, a lot can change in seven years. — https://nitter.net/GaryMarcus/status/2041953155651661977#m
-
Weak Arguments About AI: Intellectual Exercise and Critique
By
–
Dudes who won’t tag me because they know their arguments are weak sauce.*
— Gary Marcus (@GaryMarcus) 8 avril 2026
*for intellectual exercise you can list the flaws and misrepresentations in his argument below pic.twitter.com/gYqiFeUBMDDudes who won’t tag me because they know their arguments are weak sauce.* *for intellectual exercise you can list the flaws and misrepresentations in his argument below
-
Small Open Models Detect Major Cybersecurity Vulnerabilities Effectively
By
–
"But here is what we found when we tested: We took the specific vulnerabilities Anthropic showcases in their announcement, isolated the relevant code, and ran them through small, cheap, open-weights models. Those models recovered much of the same analysis. Eight out of eight models detected Mythos's flagship FreeBSD exploit, including one with only 3.6 billion active parameters costing $0.11 per million tokens. A 5.1B-active open model recovered the core chain of the 27-year-old OpenBSD bug." aisle.com/blog/ai-cybersecur…