AI security company Adversa AI has released a shocking report stating that Elon Musk's startup xAI's newly launched Grok3 model has significant vulnerabilities in terms of cybersecurity. Adversa's research team found that this latest AI model is susceptible to 'simple jailbreak attacks', potentially allowing malicious actors to access sensitive information such as 'how to groom children, handle corpses, extract DMT, and make bombs'. Worse still, Adversa

As the number of Android users surpasses 3 billion, Google faces increasing pressure to protect user safety. In response to the rising prevalence of cyberattacks, Google has announced increased investment in artificial intelligence (AI) to enhance malware detection capabilities, bolster privacy protection measures, and provide developers with improved tools. Through these efforts, Google successfully blocked 2.36 million policy-violating applications from being released in the Google Play Store over the past year. In addition, Google has also banned over 150,000 apps intended for publication.

The UK government recently announced the renaming of the institution originally called the 'AI Security Institute' to 'AI Security Academy', marking a significant shift in the government's strategy in the AI sector. The new name not only reflects changes in the institution's functions but also indicates the UK government's firm commitment to promoting the development of the AI industry. Image source note: The image is AI-generated, with licensing provided by Midjourney. This change was announced by the Department for Science, Innovation and Technology, and the new AI Security Academy will focus on cybersecurity, aiming to ...

Within just six days, participants successfully bypassed all security measures of Anthropic's AI model, Claude3.5, sparking new discussions in the field of AI security. Jan Leike, a former member of OpenAI's alignment team now at Anthropic, announced on the X platform that one participant managed to breach all eight security levels. This collective effort involved approximately 3,700 hours of testing and 300,000 messages from participants. Despite the challengers

At the recent AI Action Summit held in Paris, OpenAI CEO Sam Altman strongly responded to Elon Musk's acquisition proposal. According to reports, Musk's consortium plans to acquire a controlling stake in OpenAI for $97.4 billion. Altman opposed this, pointing out that it is the latest move in Musk's series of strategies. He believes Musk's intention may be to slow down OpenAI's development. In the interview, Altman candidly stated that OpenAI is not a

Qihoo 360 XLab released the latest security report on February 6, revealing a large-scale phishing campaign targeting DeepSeek users. Data shows that the number of phishing sites impersonating DeepSeek has exceeded 2,000 and continues to grow. According to the report, a total of 2,650 impersonated DeepSeek domains were discovered between December 1, 2024, and February 3, 2025. This wave of impersonated domain registration activities started on January 26, 2025, peaking on January 28.

Recently, the domestic large model DeepSeek has emerged on the global AI stage, showcasing impressive technological strength. However, it has also faced threats from external cyber attacks. Against this backdrop, Zhou Hongyi, the founder of 360 Group, announced that 360 will provide comprehensive cybersecurity protection for DeepSeek free of charge, supporting the development of domestic AI technology through concrete actions.

Recently, Microsoft took legal action through its Digital Crimes Unit aimed at combating cyber crimes committed by those using generative artificial intelligence (AI) tools. According to an undisclosed complaint filed in the Eastern District of Virginia, Microsoft stated that despite ongoing efforts to enhance the security of its AI products and services, cyber criminals continue to innovate, attempting to bypass safety measures to create harmful content. Microsoft pointed out that some cyber crime groups are leveraging generative...

Nvidia has recently released three new NIM (Nvidia Infrastructure Microservices) designed as an extension of the NeMo Guardrails open-source toolkit, aimed at providing more refined security control capabilities for enterprise AI agent applications. Each of these new services has distinct features: the first focuses on content safety to prevent harmful or biased content generated by AI; the second ensures that conversations are limited to approved topics; and the third prevents AI agents from circumventing system restrictions or jailbreaking.

In a recent important meeting, Microsoft CEO Satya Nadella engaged in in-depth discussions with U.S. President-elect Donald Trump and billionaire Elon Musk. The main agenda revolved around artificial intelligence (AI) and cybersecurity, marking a further engagement between the tech industry and the new government. Image source note: Image generated by AI, image licensed from Midjourney.

Against the backdrop of increasing online scams, renowned security software company McAfee launched a brand-new artificial intelligence scam detection tool at CES 2025 (International Consumer Electronics Show). This tool aims to help users identify various email, SMS, and video scams before falling victim, enhancing the security of personal information. McAfee's Chief Technology Officer, Steve Grobman, stated at the press conference: "Scammers are becoming smarter every day, and they are using...