Two separate stories mark shifting ground for model providers and platform trust. Replit's CEO confirmed OpenAI, not Anthropic, is now the default model on its platform, a visible sign that coding-tool vendors are hedging across providers rather than committing to one. Anthropic's own workforce mix is also changing, with less junior hiring and more time spent in meetings, a detail that reads as a maturing organization rather than a lean research lab. Separately, Meta quietly removed terms-of-service language that implied it could scrape third-party sites for AI training, after backlash from publishers; the company says this does not reflect a change in actual crawling behavior, which leaves the real policy unclear. Smaller items today: a Show HN tool that quizzes developers on AI-generated code before shipping, and continued scrutiny of Meta's AI products and data center PR efforts. None of this moves the technical frontier, but it sketches where competitive and reputational pressure is building.
Replit's CEO said OpenAI has replaced Anthropic as the platform's default model, though Claude remains available to customers who choose it. The same week, reporting on Anthropic's internal makeup shows a workforce that has become less junior, with more time going to meetings than before. Together these signal a market where downstream tool vendors multi-source models opportunistically, and where Anthropic itself is organizationally maturing past its early research-lab phase.
Replit's CEO says OpenAI now replaces Anthropic as the platform's default model provider, though Claude stays available to customers who opt in.
Anthropic's workforce composition and work patterns have shifted substantially over the past year, with fewer junior hires and more meeting time.
Meta quietly removed terms-of-service language that suggested it could scrape content from any domain, including news sites, to train its AI, after backlash from publishers. A Meta spokesperson said the updated terms don't reflect any actual change in how Meta crawls the web, leaving open whether the practice itself changed or just the stated policy. Two outlets covered the same removal within hours of each other, underscoring how closely publishers are now watching AI training data practices.
Meta removed terms-of-service wording that implied it could scrape third-party sites for AI training; a spokesperson says actual crawling behavior is unchanged.
After backlash, Meta updated its commercial terms again, dropping language that let it use content from any domain, including news sites, for AI training.