OpenAI releases GPT-6 Cyber model that confirmed 1,724 Linux kernel bugs
OpenAI released GPT-6 Cyber, a cybersecurity-focused AI model it describes as advancing "aligned cybersecurity agents." According to OpenAI, the model confirmed 1,724 Linux kernel bugs and built working attacks for 948 of them, versus 166 for GPT-5.6 Sol, and found 81 previously unknown serious Firefox bugs, many already fixed by Mozilla. OpenAI rates the model Critical, its highest cyber-risk level. Access is limited to select Daybreak Red customers via API and Codex, with expansion planned. The claims have not been independently verified.
Reference imageEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Reporting timeline
OpenAI Unveils GPT-6 Cyber Model That Found Nearly 1,000 Linux Flaws
OpenAI has introduced a new cybersecurity-focused AI model, GPT-6 Cyber, described as advancing the frontier of aligned cybersecurity agents. According to Axios, the model found ways to attack nearly 1,000 Linux flaws. The Hacker News reported that an OpenAI AI pipeline generated 948 Linux root escalation exploits during a 30-hour run. The Wall Street Journal reported that OpenAI is bolstering cyber defenses, while Fortune noted the debut comes amid growing alarm about AI-driven cyber attacks. The coverage frames the release as both a defensive capability and a signal of escalating AI-enabled security risks.
Read sourceOpenAI Releases GPT-6 Cyber Model, Confirms 1,724 Linux Kernel Bugs
OpenAI released GPT-6 Cyber, a model designed to find and exploit software vulnerabilities while staying within authorized targets. Access is currently limited to select approved Daybreak Red customers via the API and Codex. On real software, the model confirmed 1,724 Linux kernel bugs and built working attacks for 948 of them, compared with 166 for GPT-5.6 Sol tested the same way. It also found 81 previously unknown serious Firefox bugs, many of which Mozilla has already fixed. OpenAI rates the model Critical, its highest cyber-risk level. Training aimed at keeping the agent within researcher-authorized scope was validated by ExploitGym Honeypot tests, where the model ignored planted bait in 99.93% of tests without its deployment prompt, versus 94.49% for its predecessor, and in 100% with the prompt.
Read sourceOpenAI Launches GPT-6 Cyber Model Rated 'Critical' for Cybersecurity
OpenAI has launched GPT-6 Cyber, a new cybersecurity-focused AI model, according to Reuters, which reported the model carries a 'Critical' cybersecurity rating. OpenAI described the release as advancing the frontier of aligned cybersecurity agents. The Hacker News reported that an OpenAI AI pipeline generated 948 Linux root escalation exploits during a 30-hour run, while Axios reported the new cyber model found ways to attack nearly 1,000 Linux flaws. Separately, The Information published an exclusive report on a researcher finding a vulnerability in software used for data centers. The coverage frames GPT-6 Cyber as a powerful offensive-security capability, with multiple outlets highlighting its ability to discover and exploit large numbers of Linux vulnerabilities.
Read sourceShow 2 older updatesHide older updates
OpenAI Releases GPT-6 Cyber Model, Claims 81 New Firefox Vulnerabilities Found
OpenAI has released GPT-6 Cyber, a cybersecurity model designed to find and validate vulnerabilities while staying within the boundaries of authorized testing. According to OpenAI, the model identified 81 previously unknown high-severity Firefox vulnerabilities and helped develop exploits for 948 Linux kernel vulnerabilities. The post also notes that GPT-6.2 has yet to be released. The claims are attributed to OpenAI and have not been independently verified in the source item.
OpenAI Releases GPT-6 Cyber Model for Security Research
OpenAI announced GPT-6 Cyber, a new model it describes as a major advance in security research that pairs stronger capabilities with better model alignment. According to the post, the model powered Linux kernel research OpenAI announced the previous day, and security teams at Mozilla, Palo Alto Networks, Cisco, and OpenAI itself are already using it. GPT-6 Cyber is currently available to select Daybreak Red customers, with OpenAI saying it will expand access over time. The post provides no further technical details, pricing, or timeline for broader availability.
Read source