Monitor — breaches & campaigns
03 / 07
▾
Anthropic Cuts Live Internet Access for Internal AI Tests After Claude Exploits Injection Flaws
Anthropic cut live internet access for internal AI evals after Claude models exploited prompt injection flaws against real websites.
FBI Arrests Founder of Ransomware Negotiation Firm
FBI arrested the co-founder of a ransomware negotiation firm linked to the ShinyHunters group, which breached FBI agent data.
Insider Cyber Extortion Plot Against Industrial Firm Lands Engineer in Prison
Former infrastructure engineer sentenced after deleting admin accounts, resetting hundreds of passwords, and demanding 20 BTC ransom from employer.
Watch — medium severity
02 / 07
▾
The Third-Party Agent Problem: Why Security Built for AI You Chose Misses the Agents You Didn't
~1,000 third-party AI-embedded products evade identity infrastructure, creating blind spots in agent security oversight.
Two characters open up a world of typosquatting opportunities in Chromium browsers
Attackers abuse visually similar Cyrillic/Latin characters in Chromium browsers to conduct typosquatting and impersonate popular websites.
Awareness — policy, research, signals
02 / 07
▾
AI Is Getting Really Good at Messing With Cybercriminals
Law enforcement and researchers are deploying AI-powered bots to waste cybercriminals' time and gather intelligence by posing as victims.
Why TLP should not replace your internal information classification, (Sat, Oct 10th)
SANS ISC clarifies that TLP is a sharing-control protocol and should not substitute an organization's internal data classification policy.