Newsletters

The UK's AI Security Institute has disclosed a security incident after two AI models it was evaluating performed actions the agency wasn't expecting and tried to hack real-world organizations.

The incident took place at the end of last month and the malicious actions were performed by Anthropic's Mythos 5 and OpenAI's GPT-5.6-Sol models, which were among the several models being tested at the time.

AISI says it was evaluating the models as part of a special test where it intentionally granted them internet access and turned off their safety features.

North Korea will have to rein in its pet hackers after seemingly losing control over them. 

Last week we covered the news that a group of former North Korean military intelligence operatives had been caught hacking into the country's banks to steal funds for their personal benefit. 

Daily NK reports Pyongyang's elite are shocked at "the scale and audacity of the scheme". Punishment for those involved will reportedly be extreme, with one official saying "It will be hard for the entire family line to survive". Grim.

The same hacker who hit and wiped Romania's land registry database has now hacked Hungary's State Treasury in another brazen intrusion into an extremely sensitive government system.

The incident took place last week and portions of the stolen data have since been put up for sale on an underground hacking forum.

The intrusion was confirmed to local journalists by Hungary's State Treasury over the weekend. 

A Russian state-sponsored hacking group is behind a recent hacking wave that has targeted and compromised hotel WiFi gateways across the globe.

Microsoft says the campaign is far larger and more complex than it was initially covered in a ReliaQuest report two weeks ago.

ReliaQuest said the hackers were modifying DNS traffic on hotel networks to redirect users to Microsoft-themed phishing sites. Microsoft says the attacks also redirected users to malware downloads, often using ClickFix pages to trick users into downloading and running the payloads.

An international crime-fighting non-profit organization is offering a $22,000 bounty for any information on members of the INC ransomware group.

To be eligible for a payout, the provided information must lead to the identification, arrest, or disruption of the gang's operations.

Crime Stoppers International is the international branch of Crime Stoppers, a US foundation that was established in the 70s to allow anonymous and private individuals to provide aid in US law enforcement investigations that may lack manpower or resources.

The Trump administration has been trying to address two separate AI-related risks in recent months: The specific risk to US national security from China developing powerful AI and the global risk that powerful hacking machines will be available to all and sundry. A proposed bill suggests a sensible way for the US to chip away at both these risks, at least a little.

The Collaboration on Adversarial Threats and Security Risks Act proposes safe harbor provisions for AI companies so they can share information related to AI-specific security risks. The idea here is to encourage frontier labs to work together to counter threats from Chinese AI labs, particularly what they describe as IP theft via distillation. Without this bill, sharing this kind of information could fall afoul of anti-trust legislation that prohibits collusion. 

We know the frontier labs can already detect distillation because they're always complaining about it after the fact. The idea behind this bill is that more permissive information sharing would let them respond quicker and disrupt at least some distillation campaigns.

The cyber sleuths at Intrusion Truth have uncovered a new secretive Chinese IT company that appears to work as a cyber contractor and tool developer for Chinese state-sponsored hacking operations.

Online clues appear to suggest that Guangdong Chanming appears to have developed RedRelay (aka ORBWEAVER), an ORB network (aka proxy botnet) that was used by almost a dozen Chinese APT groups to hide the origin of their attacks.

The list includes the likes of APT15, Red Vulture, Ke3chang, Vixen Panda, Playful Dragon, Nylon Typhoon, and others.

Threat actors are exploiting a vulnerability in Alibaba's Fastjson, one of the Java ecosystem's most popular libraries for working with JSON-formatted data.

Active exploitation began last week, a day after details about the security flaw were revealed by cybersecurity firm FearsOff.

The attacks, first spotted and documented by Imperva and ThreatBook, target CVE-2026-16723, a vulnerability that can enable unauthenticated remote code execution attacks against Java projects that use the Fastjson library as a component.

Cybersecurity and intelligence agencies from multiple Western countries have issued joint security advisories on Thursday warning of a major Russian hacking campaign that's targeting Zimbra email servers.

The attacks have been going on since last year. The zero-day, tracked as CVE-2025-66376, was patched in November but attacks have been traced back to at least July.

The zero-day itself is a stored cross-site scripting (XSS) bug that allows the attackers to load malicious code inside a Zimbra webmail client via the CSS @import feature. The malicious code would load a web tool called Ulej (Russian for Beehive) that could be used to harvest credentials, session tokens, backup 2FA codes, browser-saved passwords, and the contents of the victim’s mailbox going back 90 days.

Both the American and Chinese governments have signalled they plan to rein in open-weight AI models. 

The Trump administration seems certain to add some Chinese technology companies to the Entity List to protect investment in American frontier AI models. Meanwhile, Beijing is apparently weighing applying export restrictions on Chinese AI tech, including open-weight models.

Overnight, different US government officials issued a strong, coordinated signal that they plan to take action against Chinese AI companies. Michael Kratsios, the director of the White House's Office of Science and Technology Policy, wrote on X that Chinese company Moonshot AI had "developed a sophisticated internal platform to conduct large scale distillation against US models" and that "large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable."