Open AI Coalition Gains Support Amid Security Concerns

After a Hugging Face breach where closed security tools blocked forensic analysis, open-weight GLM 5.2 was deployed to contain it. Perplexity announced it is joining OSAA to support open tools for security. All major AI labs—Google, Meta, and OpenAI—have signed the coalition's joint letter in support of open AI; Anthropic remains the only holdout.
Key insights
Hugging Face Breach Validates Open-Weight Models as Essential Security Tools
When Hugging Face was breached, closed security tools could not distinguish attackers from defenders and blocked forensic analysis. The response team turned to running open-weight GLM 5.2 on their own infrastructure to contain the incident. Perplexity, citing this case, announced its membership in OSAA to promote open tools for security, arguing that openness is critical for effective incident response.
Industry Giants Rally Behind Open AI Coalition, Excluding Anthropic
Google, Meta, and OpenAI have all signed the coalition's joint letter supporting open AI. Anthropic has refused to sign, making it the only major model laboratory not endorsing the group. The coalition appears to be gaining broad industry backing, with the notable exception of one prominent AI safety-focused company.