Anthropic is hiring fewer juniors while websites bar AI agents and Meta quietly cuts scraping language, with alignment theorists arguing the field needs deeper theory rather than trial and error.
-
Anthropic hires fewer juniors and spends more time in meetings
- The share of junior-level employees in technical departments fell from 21% to 14% since early 2025.
- The share of junior technical hires fell from 27% to 9%.
- Technical teams fell from 61% to 56% of the overall workforce.
- Meeting time per employee is up 20% compared to a year ago, and up 16% for technical contributors.
Why it matters. The report is a rare look at how an AI company organizes its own staff and work. It suggests human attention and judgement may become the bottleneck as AI output grows. For developers, it signals that junior hiring may shrink while review and coordination take more time.
-
Websites are closing their doors to AI agents
- Users of agents such as Meta's Muse, xAI's Grok, OpenAI's Dots and Instinct report blocks on top websites.
- Yelp says it only allows AI agents that pay for its data licensing program, and Delta Air Lines prevents them from booking tickets.
- JPMorgan Chase and Charles Schwab confirmed they block AI agents from accessing accounts due to security concerns.
- Amazon said Muse's continued access violates its site policies, and Amazon previously sued Perplexity over its shopping agent.
- Anti-bot systems also block some agent activity, and Cloudflare now blocks crawlers that scrape for both search and AI training.
Why it matters. Agents only help if they can reach the sites people use. Fewer places where agents work means fewer useful tasks they can do, according to one AI safety evaluator. Developers building agents will need to handle blocks, anti-bot checks and site rules.
-
Meta removes terms language that raised scraping concerns
- Meta removed language from its commercial terms that suggested it could use content from sites, including news sites, to train its AI.
- Experts said the provision could let Meta scrape publisher sites with a business account and override robots.txt blocking instructions.
- Meta had earlier said there was no change to how it uses web crawlers, and that any suggestion to the contrary was false.
- Meta now says content owners can still use robots.txt to signal their preferences.
Why it matters. The episode shows how AI training rights over web content are being contested. Developers who build crawlers or publish content should check how robots.txt signals are respected. The terms still matter for how companies describe their crawling.
-
A new picture of AI corrigibility built on respect for authority
- The author argues that existing definitions of corrigibility are deficient in one way or another.
- The article says the fundamental concept underlying corrigibility is respect for authority, using a healthcare proxy as an intuitive guide.
- The framework first restrains the agent from overstepping or interfering with reserved human authority, then has it act on the principal's behalf as the principal would judge best.
- The article presents an initial formalism and proves theorem-level guarantees in three increasingly realistic formal settings.
Why it matters. Corrigibility is central to theoretical AI alignment, and the article says no one can agree on what it means. The framework aims to match the original 2015 desiderata and to support the formal safety guarantees the concept needs. Developers working on powerful AI systems would find the formal settings relevant to how such goals could be specified.
-
Kosoy argues AI alignment needs deep theory, not trial and error
- Vanessa Kosoy defines the technical AI alignment problem as designing an AI system that brings superhuman capability to multi-domain, long-term planning while staying aligned with user intentions and preferences.
- She argues that the minimal task for safe AI is creating a reliable global defense system against dangerous AI.
- She argues that the incremental approach used in the AI industry fails if the problems lead to AI takeover.
- She says aligned AI needs aligned preferences, beliefs and decisions, and that philosophical questions about these are not easily brushed aside.
Why it matters. The essay frames alignment as a requirement for AI that can reason about the whole planet and plan for decades or centuries. It argues that finding problems after deployment is too late if the problems lead to takeover. For developers, it means trust depends on beliefs and decisions, not only on goals.
