{"id":49473,"date":"2026-07-30T12:24:31","date_gmt":"2026-07-30T19:24:31","guid":{"rendered":"https:\/\/occupysf.net\/?p=49473"},"modified":"2026-07-30T12:24:32","modified_gmt":"2026-07-30T19:24:32","slug":"casar-demands-hearing-after-openai-model-goes-full-cybercriminal-on-another-firm","status":"publish","type":"post","link":"https:\/\/occupysf.net\/index.php\/2026\/07\/30\/casar-demands-hearing-after-openai-model-goes-full-cybercriminal-on-another-firm\/","title":{"rendered":"Casar Demands Hearing After OpenAI Model Goes \u2018Full Cybercriminal\u2019 on Another Firm"},"content":{"rendered":"\n
\"Congressional<\/figure>\n\n\n\n

Congressional Progressive Caucus Chair Greg Casar (D-Texas) speaks during a July 16, 2025 press conference on Capitol Hill in Washington, DC.<\/p>\n\n\n\n

\u00a0(Photo by Bill Clark\/CQ-Roll Call, Inc. via Getty Images)<\/p>\n\n\n\n

\u201cAI systems are creating real-world consequences,\u201d said one expert. \u201cCongress should treat this with the urgency it deserves.\u201d<\/h2>\n\n\n\n

Brett Wilkins<\/a><\/p>\n\n\n\n

Jul 29, 2026 (CommonDreams.org)<\/p>\n\n\n\n

US Rep. Greg Casar<\/a> on Wednesday called for a congressional hearing following reporting that the OpenAI artificial intelligence<\/a> model that autonomously breached the systems of the open-source platform Hugging Face during recent internal testing also independently hacked another tech company.<\/p>\n\n\n\n

Akshat Bubna, chief technology<\/a> officer at New York-based Modal Labs, told<\/a> Reuters that OpenAI\u2019s model exploited vulnerable code written by a customer that was hosted on the company\u2019s platform.<\/p>\n\n\n\n

RECOMMENDED…<\/p>\n\n\n\n

\"Open<\/a><\/figure>\n\n\n\n

Calls for Guardrails Grow as OpenAI Discloses \u2018Unprecedented\u2019 Autonomous Cyber Attack<\/a><\/h2>\n\n\n\n
\"Who<\/a><\/figure>\n\n\n\n

Who Are the Biggest AI Villains? An Industry Watchdog Has Made a List<\/a><\/h2>\n\n\n\n

Last week, calls for stronger AI guardrails mounted<\/a> after OpenAI CEO Sam Altman admitted to a \u201csignificant security incident\u201d during evaluation of one of the ChatGPT maker\u2019s models.<\/p>\n\n\n\n

The target of that breach, the popular open-source platform Hugging Face, said Tuesday that the OpenAI model\u2014widely described in the press as \u201crogue\u201d\u2014broke into a sandbox<\/a> \u201chosted on a third-party provider\u2019s infrastructure<\/a>\u201d and used it to launch a broader attack.<\/p>\n\n\n\n

Bubna stressed that Modal\u2019s platform was \u201cnot compromised in any way.\u201d<\/p>\n\n\n\n

Experts underscored that OpenAI\u2019s purportedly \u201crogue\u201d model was, in fact, performing a specific assigned task, which involved executing complex cyberattacks and exploiting vulnerabilities in controlled scenarios.<\/p>\n\n\n\n

However, AI specialists said the incident illustrated the problem of misalignment. As artificial intelligence advances to the point where it will very likely outsmart humans, the challenge of ensuring that advanced systems reliably pursue goals that match what humans actually want becomes increasingly difficult\u2014and dangerous to the point where pioneers and preeminent experts in the field,\u00a0including Altman<\/a>, believe there is a\u00a0growing chance<\/a>\u00a0that AI could one day cause human\u00a0extinction<\/a>.<\/p>\n\n\n\n

\n

OpenAI\u2019s rogue AI agent went full cybercriminal

It escaped during testing, spent 4.5 days hacking, and compromised four separate services, including Hugging Face and a Modal Labs customer.

OpenAI only found out after the damage was done\u2026 and the FBI had already been alerted.\u2026<\/p>— Mario Nawfal (@MarioNawfal) July 29, 2026<\/a><\/blockquote>