Security Bulletin

The new AI security controls follow the Hugging Face incident last month, though many of these additions perhaps should have been in place prior to the frontier models escaping.

US Bank confirmed it is aware of the claims and is investigating, but declined to provide specifics on communication with the extortionists or the ransom amount.

A browser-based variant of NASA's AMMOS Instrument Toolkit (AIT), specifically versions up to 2.5.1, has a critical vulnerability that could allow an unauthenticated attacker to issue commands to spacecraft and instruments, and potentially execute...

The coming threat of super-powerful computers capable of cracking today’s algorithms requires upgrading encryption now. Tech companies have begun building defenses.

Over 50,000 Stripe API keys have been exposed across public code repositories, GitHub Actions logs, and misconfigured web servers, demonstrating the immediate risks of leaked credentials.

The AI Workforce Consortium's report highlights that AI skills are now present in 28.5% of G7 cybersecurity job postings, a substantial increase from 14.2% a year prior.

The vulnerability arises from a discrepancy between two loops within the File Upload module: one for validation and one for processing.

The report defines the AI sector broadly, encompassing organizations, technologies, and industries involved in the development, training, deployment, and operation of AI systems, including frontier models, data centers, and AI-specific hardware.

The attack, developed by Rony Utevsky, a researcher at Adversa, is dubbed "cryptographic context injection" and exploits a fundamental weakness in how AI models process information.

Socket researchers identified 40 malicious extensions and 37 others disguised as unrelated utilities, all linked through shared code, infrastructure, and publishing artifacts.

Discovered by Zimperium's zLabs team, ToxicPanda 2.0 leverages the Android Accessibility Service to enable wireless debugging, effectively gaining shell access and bypassing standard runtime consent prompts.

Logitech's research reveals that 98.8% of IT decision-makers believe gaming directly impacts their job performance.