Anthropic has suspended its new AI model, Claude Fable 5, after US authorities raised security concerns. The company stated it was ordered to suspend foreign nationals from using the model, which it described as "too powerful." The suspension comes amid a separate lawsuit between Anthropic and the Trump administration over AI tool usage.
France's cybersecurity agency Viginum has accused Israeli tech company BlackCore of orchestrating a digital interference campaign targeting Scottish First Minister John Swinney and the Scottish government during recent elections. The firm allegedly used proxy social media accounts to spread disinformation, with similar operations noted in France, New York, Togo, and Angola. The sponsor behind BlackCore's actions remains unidentified.
China's State Administration for Market Regulation, Cyberspace Administration, and National Railway Administration have jointly summoned seven online travel platforms, including Ctrip and Qunar, to address irregular operations such as misleading ticket sales and improper collection of user data. The platforms are urged to comply with relevant laws and regulations to maintain market order and protect consumer rights.
Russia has lifted its ban on the gaming platform Roblox following widespread complaints from children and parents. The platform was banned in December for allegedly distributing extremist materials and promoting "LGBT propaganda."
CISA issued a new directive requiring federal agencies to patch software vulnerabilities within three days for critical cases, driven by AI's potential to accelerate exploitation. The rule supersedes previous mandates and aims to improve prioritization amidst funding challenges.
Anthropic's new Fable AI model, a public version of its cybersecurity tool Mythos, is facing criticism from researchers for overly strict guardrails that block even benign requests related to cybersecurity or biology.
China's cybersecurity watchdog warns against third-party AI 'skills' packages that bypass safety guardrails or offer crypto-mining functions, citing risks of data leaks and money laundering. Users are advised to use official channels and limit permissions.
Adaptive Security has partnered with comedian Conan O'Brien for a 15-part educational video series aimed at training clients and employees on cybersecurity threats like phishing and deepfakes, addressing the common issue of boring corporate training.
Anthropic launched Fable 5, its advanced AI model, restricting sensitive uses like cybersecurity and biology/chemistry queries to a lower-tier model. An unrestricted version, Claude Mythos 5, is available to select partners, including US government entities, via Project Glasswing.
Anthropic launched Claude Fable 5, a new AI model with enhanced capabilities but restricted access to sensitive topics like cybersecurity and biology. The company aims to prevent misuse by malicious actors, though this may limit harmless requests.
Researchers have developed a proof-of-concept AI-powered computer worm that can autonomously identify vulnerabilities, generate tailored attack strategies, and spread across networks, adapting its tactics in real-time without human operators.
Anthropic has launched Claude Mythos 5, an AI model with advanced cybersecurity capabilities, initially restricted to approved organizations. They also released Claude Fable 5 for broader public use, featuring similar capabilities with enhanced safeguards.
Google's cybersecurity teams expose UNC3753's financially motivated data theft campaign targeting US professional, legal, and financial services via vishing and social engineering, with a 3-day ransom deadline.
Secondo Kaspersky, il danno medio per utente in Italia a causa di attacchi cibernetici è di 770 euro. Polizia Postale, Csirt e Cert-Agid rilevano frequenti tentativi di sfruttamento.
A former IBM VP claims the company was hacked by foreign governments (including China's APT 10) multiple times between 2013-2018, covering up breaches despite being a major U.S. federal cybersecurity vendor.
Anthropic sends engineers to NSA to assist in utilizing Mythos AI for cybersecurity, amid a federal ban and concerns over potential misuse.
Several Indian organizations in cyber, telecom, finance, and banking have received early access to Anthropic's advanced Claude Mythos AI model. The Indian government is actively working to expand this access, with key public entities like CERT-In set to receive previews to bolster cybersecurity for critical infrastructure.
CrowdStrike exceeded Q1 earnings and revenue expectations but saw its stock fall 10% post-report. The cybersecurity firm announced a 4-for-1 stock split and highlighted growth driven by AI adoption and new security solutions.
Moldova is moving to sever economic ties with Russia, with a Russian envoy criticizing the leadership's European integration course. Meanwhile, Russia is ready to normalize relations with the US if its interests are respected, and has imposed retaliatory sanctions on five UK citizens. Ukraine continues drone attacks on Russian regions, causing casualties.
Anthropic's AI cybersecurity model, Mythos, is now accessible to India and ~150 organizations globally via Project Glasswing. This inclusion, unusual for a non-allied power, highlights India's strategic importance in AI and cybersecurity.
Microsoft's Build conference showcased a new AI-first strategy, featuring in-house reasoning models like MAI-Thinking-1, a cybersecurity tool (MDASH), and 'Autopilot' AI agents integrated into a Copilot super app. This marks a significant pivot from its reliance on OpenAI, as Microsoft aims to compete directly with top AI labs.
Qi Andong, Chairman of Qi Anxin Group, stated at the 2026 Global Digital Economy Conference that the AI revolution is beneficial for the cybersecurity industry, with innovative leading security firms being the biggest beneficiaries. He highlighted new security demands in AI vulnerabilities, data security, and full-stack AI application security.
IIT Roorkee acknowledged a cloud configuration issue affecting JEE Advanced 2026 infrastructure after a researcher claimed candidate data was exposed. The institute stated data was read-only and no alteration was possible, while initiating corrective actions.
President Trump signed an executive order for early government access to advanced AI models to assess cybersecurity risks and protect critical infrastructure. The voluntary framework allows US agencies to vet AI systems up to 30 days before public release.
