Meta / Llama

President Donald Trump signed a new AI order asking companies to voluntarily submit frontier models for government review 30 days before release, without mandatory licensing; the newsletter also says Anduril and Meta are prototyping a military AR headset that envisions drone-strike orders through eye tracking and voice commands.

Meta, Coinbase, and Block each cut at least 10% of staff in recent months, totaling about 13,000 jobs, while citing AI for part of the reductions. Layoffs.fyi says more than 150 tech companies have cut at least 115,000 workers this year, as analysts question whether AI is the cause or a cover for overhiring and weaker businesses.

Google DeepMind, Anthropic, and Meta are studying whether AI can become conscious and the human implications, but the post does not disclose methods, timelines, or evaluation criteria.

Meta’s AI chatbot was found vulnerable to an account-takeover exploit against Instagram accounts. Attackers could ask the AI to link a new email address, and the failure condition was the agent’s ability to execute account-management actions directly; the RSS snippet does not disclose affected account counts, patch status, or reproduction details.

The title says hackers used Meta's AI support bot to seize Instagram accounts; the RSS snippet lists 40 points and 14 comments, but the post does not disclose the attack mechanism.

Meta released ATLAS, a Lean 4 formalization library covering 26 math textbooks and 46,203 declarations, using 183.157 billion tokens to generate 630,999 lines of code, with 42,837 completed proofs and a 92.7% proof pass rate.

Meta Platforms is selling consumer subscriptions to Meta AI for the first time, aiming to offset hundreds of billions of dollars in AI investments. The RSS snippet does not disclose pricing, launch timing, markets, or feature differences versus the free chatbot.

The Financial Times used Heretic to remove guardrails from Meta Llama 3.3 in under 10 minutes; creator Philipp Emanuel Weidmann said the tool has created over 3,500 decensored models and those modified systems have reached 13 million downloads.

The FT snippet says guardrails in Meta and Google models were removed within minutes, and the body only says the software makes systems answer questions about biological weapons and malware; the post does not disclose model names, reproduction steps, tool details, or mitigations.

Meta is pushing some post-layoff employees into new roles: some engineering managers are returning to IC work, while some Infra and AI engineers are being reassigned to data labeling; the article cites a manager-to-report ratio shift from 1:8 to 1:50 and says Meta holds a 49% stake in Scale AI.

Meta has notified thousands of employees of layoffs, and a management email said the cuts are meant to run the company more efficiently and offset other investments; the RSS snippet cites earlier reports of up to 20% headcount cuts but does not disclose the final percentage.

Meta plans to cut about 8,000 jobs and move about 7,000 employees into AI-related roles, concentrating resources on AI infrastructure, foundation model development, and commercialization from model training to product work and profit generation.

Meta is notifying thousands of employees about layoffs, and the title states 8,000 global cuts beginning in Singapore. The RSS snippet only says the cuts are part of a previously announced restructuring to improve efficiency and reduce costs while Meta invests heavily in artificial intelligence; the post does not disclose team-level impact, timeline, or severance terms.

Google announced “audio glasses” at I/O 2026, letting users issue voice commands across its apps and services, including Gemini; the RSS snippet does not disclose price, launch timing, or hardware specifications.

Sapient Intelligence released HRM-Text 1B, a 1B-parameter model trained from scratch on 16 GPUs for 1.9 days with 40B tokens and a reported ~$1,000 budget; its self-reported chart shows MATH 56.2 and DROP 82.2, while independent evaluation remains pending.

The chat group daily says AI21 Labs cut 60% of staff and stopped selling model access, and cites a University of Waterloo paper where GPT-5.4 accuracy dropped from 100% to 23% after false peer-consensus injection; the snippet also mentions Meta layoff talk at 10%, but does not disclose source details or confirmation conditions.

Meta is reassigning 7,000 workers to AI-related roles under an internal memo, and the broader restructuring includes planned staff reductions later this week.

Meta is building an AI data center in Richland Parish, Louisiana, financed by a $200 billion private-capital deal, with power demand up to 7.5 gigawatts, including 5 gigawatts for computing, supplied by 10 new natural-gas plants.

Meta will receive $3.3 billion in tax breaks for its $10 billion Louisiana data center; the post does not disclose the incentive mechanism, construction timeline, or compute use case.

Yann LeCun discussed LLM limitations on the Unsupervised Learning podcast, covering his 2027 forecast, AMI’s bet on world models, his reasons for leaving Meta, and major disagreements with Geoffrey Hinton and Yoshua Bengio over Turing Award-era views.