The Hacker News
OpenAI addresses GPT-6.2 jailbreak flaw after UK researchers' report
Part of the 8-Digits Team story · since 19 Nov
- Thu 19 Nov, 20:30 UTC. The Hacker News, first report. OpenAI addresses GPT-6.2 jailbreak flaw after UK researchers' report. This article, 1st of 1.
The ThreatsDay bulletin for 19 November 2026 highlights a mix of new AI-enabled attack capabilities, supply-chain and third‑party breaches, and routine software fixes, all underscoring how commonplace tools and services can be weaponised. Notably, OpenAI announced GPT‑6.2 Astra and Sol with claimed cyber capabilities including vulnerability discovery and patching, while UK researchers reported a universal jailbreak against pre‑release GPT‑6.2 Astra, prompting OpenAI to address a regression in jailbreak robustness.
The report also notes a related push by third parties to embed independent evaluators to test model alignment and safeguards.
Among concrete incidents, WordPress released version 7.1.5 with multiple security fixes, and Ireland’s Data Protection Commission fined Match Group €11 million for Tinder data-handling failures. There is a significant ransomware footprint as well: a Kazakh national associated with the 8‑Digits Team extradited to the U.S. over alleged extortion of schools, hospitals and other organisations, with more than $500 million in ransom demands linked to the operation.
Other notable items include BubbleLoader, a PNG‑based malware delivery technique that patches host applications to inject malicious payloads, and GravityRAT campaign targeting Indian defence job seekers via a fake recruitment portal.
The roundup also covers a range of additional risks: a 52‑bug set in PostgreSQL, a critical pre‑auth MongoDB flaw (CVE‑2026‑118383) capable of crashing or enabling RCE, and a broader spectrum of data‑exfiltration, credential‑phishing and supply‑chain threats tied to infrastructure and cloud services. The piece attributes each item to its respective researchers, vendors or outlets and cautions readers that some campaigns are only documented indicators of compromise rather than confirmed exploitation. SOURCE_UNAVAILABLE
How the story developed
1 article from 1 outlet since 19 Nov. Still developing; the latest report was 33 minutes ago. This was the first report.
- Thu 19 Nov, 20:30 UTC. The Hacker News, first report. OpenAI addresses GPT-6.2 jailbreak flaw after UK researchers' report. This article, 1st of 1.
- No CVE
- Larger: an outlet’s first report
- Coverage over time
Show all 1 articles as a listHide the list
- Thu 19 NovDay 01
- OpenAI addresses GPT-6.2 jailbreak flaw after UK researchers' report20:30 · The Hacker News · First report · This article, 1st of 1