OpenClaw Releases 2026.9.5 with 64 Commits and 4,179 PRsClaude Code v2.1.278 Defaults to Server-Side Classifier for API Usersdatasette-auth-github 1.0 Released, Fixes Session Expiry Bug6 Clones of Jev Appear Within 2 DaysOpenAI and Anthropic Accused of Overselling AI Security BreachesLawsuit Alleges Anthropic, OpenAI, Others Made Illegal AI Slowdown AgreementMicrosoft Director Calls AI Scraping 'Largest Theft of Labor in Human History'Opinion: You Should Almost Never Use AI to WriteOpenAI Codex Releases 0.156.0-alpha.9OpenClaw Releases 2026.9.5 with 64 Commits and 4,179 PRs

Claude Code released v2.1.278. The update changes auto mode for Claude API and Enterprise users, as well as on Bedrock, Vertex, Foundry, and gateways, to default to the server-side classifier, which does not charge for classifier overhead. Users can opt out on Bedrock, Vertex, Foundry, and gateways by setting CLAUDECODEAUTOMODESERVER=0. This change reduces auto mode costs for enterprise users.

Simon Willison released datasette-auth-github 1.0. While running the GitHub login plugin on the agent.datasette.io demo site, he noticed authenticated sessions weren’t lasting long because the plugin set cookies without a Max-Age parameter, causing them to expire when the browser closed. Version 1.0 fixes this, allowing login sessions to persist properly. Developers using this plugin can upgrade to prevent users from being repeatedly logged out.

Six clones of a project called Jev emerged within just two days, according to Latent Space’s AINews. The newsletter quipped that ‘imitation is the sincerest form of flattery.’ The rapid replication highlights how quickly AI tools and models can be copied once released, suggesting that first-mover advantages in AI are shrinking. Developers may need faster iteration cycles and stronger differentiation to stay competitive.

According to the New York Post, insiders claim OpenAI and Anthropic oversold AI security breaches to pressure federal regulators into protecting their turf. The story sparked discussion on Hacker News with 33 points and 17 comments. If true, the allegation could undermine public trust in AI companies’ safety claims and draw scrutiny to how AI safety regulations are shaped.

According to the Associated Press, an antitrust lawsuit alleges that Anthropic, OpenAI, SpaceXAI, and Google made an illegal agreement regarding the pace of AI development. The lawsuit drew attention on Hacker News with 43 points and 14 comments. If the allegations hold, these companies could face antitrust penalties. The case could affect the competitive landscape and regulatory direction of the AI industry.

According to Tom’s Hardware, a Microsoft director called AI scraping ‘the largest theft of labor in human history,’ while an OpenAI head branded ChatGPT an ‘existential threat’ to publishers. These statements come from legal briefs filed in the New York Times lawsuit. The report sparked discussion on Hacker News. These internal remarks reflect growing tensions between AI companies and content creators.

Eric H. Grunewald published an article on Substack arguing that people should almost never use AI to write. The piece garnered 202 points and 115 comments on Hacker News, sparking widespread discussion. The author contends that AI writing weakens personal expression and depth of thinking. Amid the proliferation of AI writing tools, this view represents some creators’ concerns about AI replacing human writing.

OpenAI Codex released version 0.156.0-alpha.9, following a series of rapid iterations from 0.156.0-alpha.6 through alpha.8. The release remains in alpha testing with no detailed changelog disclosed. The frequent alpha iterations suggest the Codex team is actively developing and testing new features. Developers should watch for a stable release.