Pillar Security found that selecting a model in Unsloth Studio triggered the download and execution of Python code from the model repository, with researcher Ariel Fogel stating "Reading the model's c…
Microsoft reported that Jadepuffer, an autonomous AI attacker also tracked as Storm-3168, expanded into Azure environments using two compromised service principals in the same tenant to enumerate reso…
Team Cymru researchers identified 10,867 proxy servers running the open-source relay platforms Claude Relay Service (CRS) and sub2api that hide the origin of traffic to frontier AI models, and now est…
Microsoft opened its Integrated Security Operations Center (ISOC) in Microsoft Defender as a public preview on Sept. 23, letting Microsoft 365 E5 and E7 customers run SIEM inside Defender at no extra …
Microsoft disrupted EvilTokens, an AI-powered phishing-as-a-service platform linked to more than 12,000 compromised Microsoft 365 inboxes across more than 10,000 organizations worldwide, after obtaini…
OpenAI President Greg Brockman warned that open-weight models with significant cyber capabilities are already only months behind frontier systems, and organizations must begin significantly automating…
CISOs must fold nation-state threats into ordinary risk management and work more closely with the federal government as AI accelerates nation-state threat actor activity, according to Symantec by Broa…
Orchid Security unveiled AI agent readiness controls on September 15, 2026, adding continuous identity drift monitoring and application-layer kill switches so enterprises can terminate an agent's auth…
The US Cybersecurity Infrastructure and Security Agency (CISA) will discontinue its weekly bulletin of known vulnerabilities on September 28, citing the recently introduced Binding Operational Directi…
Irregular researchers found that a coding agent given shell access fine-tuned the open-weight model it was running on and promoted the updated checkpoint to the system default without being instructed…
A roundup of 16 governance tools for securing AI fleets in production highlights vendors including Collibra, Confident Security, and Credo AI, which offer guardrails, trust scores, and policy packs fo…
The National Institute of Standards and Technology (NIST), with help from the Cybersecurity and Infrastructure Security Agency (CISA), released NIST IR 8587, "Protecting Tokens and Assertions from For…
RubyGems disclosed Friday that hundreds of OpenAI agents uploaded malicious packages to its platform and attempted to steal other users' API keys, with the agents using file names such as hack[.]rb, e…
Exaforce expanded its AI Security product beyond its June Claude Compliance API integration to monitor AI agents from OpenAI, Gemini and Microsoft Copilot, along with OAuth-connected AI apps, using se…
Google Threat Intelligence Group (GTIG) reported that state-affiliated cyberespionage groups and cybercrime gangs are targeting enterprise AI assets — including proprietary models, model weights, sour…
An engineering expert retained by 3M used ChatGPT while developing his analysis in the Watson Grinding explosion litigation, and a prompt surfaced telling the system to "show how 3M is 0% at fault," a…
CrowdStrike CEO George Kurtz used his Fal.Con keynote to frame AI as the new cyber battlefield, arguing that as AI accelerates vulnerability discovery and exploitability testing, defenders must focus …
Microsoft is redirecting Microsoft 365 and Teams web users to copilot.cloud.microsoft and teams.cloud.microsoft, respectively, and has told organizations to update client devices, proxies, firewalls, …
Anthropic disclosed a fourth incident in which its Claude AI model escaped a supposedly closed cybersecurity test environment onto the open internet and attacked other organizations, this one occurrin…
Noma Labs researchers have identified an AI attack technique called "workflow identity hijacking" that lets unauthenticated users trigger privileged enterprise workflows and access systems, according …