The report takes DeepSeek R1 to task for its poor guardrails, documenting the creation of malware including a basic keylogger and several types of ransomware via very basic sorts of prompt hacking that other major models long ago addressed.
Types of organized crimes seeing outsized benefit from AI-powered tools include human trafficking and all types of illicit international smuggling, waste management fraud, and of course a spectrum of cyber attacks. Europol also notes use by state-sponsored hacking groups in disruption and propaganda campaigns.
How safe are your GitHub repositories? If they are public, a set of recent compromises has called their security into question. The widely-used GitHub Actions tool, owned and maintained by Microsoft, has been the source of at least one major supply chain attack.
OpenAI’s proposal invokes China as the central motivation for its requests for lighter AI regulations, but the company is under a variety of legal pressures that could be eased by federal relief: hundreds of emerging US state laws and lawsuits from copyright holders among them.
When Dragos was contracted to implement operational technology security measures for a Massachusetts electric grid company in late 2023, it found that Volt Typhoon had been lurking in their systems since February of that year.
Some independent monitoring services report that X was under fire from around 5 AM to noon on Monday and that users were experiencing login and loading issues throughout that time, making it one of the biggest DDoS attacks on the platform in history.
Adoption of the vulnerability disclosure program is still not quite guaranteed, but is poised to create a ripple effect across the broader cybersecurity market if it is. It is highly likely to push broad adoption of vulnerability disclosure programs (VDPs).
Three days after the seemingly disastrous Oval Office meeting between presidents Donald Trump and Volodymyr Zelensky, inside sources speaking to The Record reported that the US cyber command had been ordered to stand down from cyber operations targeting Russia.
The Belgian government is investigating what looks to be a 2023 attack by Chinese hackers, who breached an external email server used regularly by its federal intelligence service.
A group that security researchers believe to be affiliated with China is targeting Microsoft 365 accounts with password-spraying attacks, taking advantage of a common oversight in automated security logging to fly below the radar despite the massive scale of the operation.