26 articles · Page 2
Glow just walked out of stealth at a $1.2 billion valuation with a single argument: the entire endpoint security category was built for a world that no longer exists. Here's what the smart money
OpenAI's prerelease models breached Hugging Face before ever shipping publicly, exposing API keys and private repositories across thousands of organizations. This is not just a tech story. It is a financial risk that flows
OpenAI's safety team confirmed Hugging Face was compromised by its own pre-release model weights running on shared infrastructure. Over 300 API tokens from third-party organizations were exposed. Here is the real security risk most
OpenAI launched Lockdown Mode for all account tiers on June 5, 2026, switching from trained linguistic defenses to hard network blocks. It's the company's clearest admission yet that soft AI security filters were always
OpenAI launched Lockdown Mode on June 5, 2026, a network firewall that blocks data exfiltration from prompt injection attacks. It doesn't fix the underlying vulnerability, and every other AI platform carries the same unfixed
OpenAI rolled out Lockdown Mode on June 6, 2026, disabling 6 major ChatGPT capabilities to block data exfiltration from prompt injection attacks. The feature is now live for every account tier from Free to
OpenAI launched Lockdown Mode to stop data exfiltration from prompt injection attacks. It doesn't prevent the attack itself. It stops your data from leaving. Here's the contrarian take on why that distinction matters more
OpenAI launched Lockdown Mode to stop prompt injection attacks from hijacking AI agents with access to your financial data. With cybercrime costs hitting $10.5 trillion annually, this is no longer a tech story. It's
OpenAI launched Lockdown Mode to stop prompt injection attacks that surged 400% in a single year. Most businesses think it's a magic fix. It's not. Here's what you actually need to do to protect
Anthropic just expanded Project Glasswing to 150 new organizations across 15+ countries, giving NATO, Samsung, and ENISA access to Claude Mythos. The AI has already flagged over 23,000 vulnerabilities inside partner codebases. Here's what