AI Safety stories - Page 22

Elon Musk's xAI launches Grok 4, a powerful new AI model now available for users and businesses with advanced multitasking and multimodal features.

Google's Veo 3 AI model creates HD videos with realistic audio from text prompts, revolutionising video content creation for creatives and businesses alike.

Matt Johnson of Temus highlights Singapore's AI success lies in strategic, cautious deployment and real-world learning, beyond mere prompting hype.

Genesys launches Cloud AI Studio with AI Guides, enabling firms to build scalable, ethical virtual agents for complex customer interactions without coding.

A Cycode study reveals cautious yet growing adoption of agentic AI in application security, with 60% early adopters reporting improved productivity and risk mitigation.

The Cloud Security Alliance has launched the AI Trustworthy Pledge to promote responsible, transparent, and ethical AI development and usage globally.

Guardian agent technologies are set to capture 15% of the AI market by 2030, ensuring secure and trustworthy AI interactions, says Gartner.

Google Cloud launches Imagen 4, Veo 3, and Lyria 2 on Vertex AI, boosting image, video, and music generation with advanced AI media models.

Google DeepMind has unveiled a new strategy to bolster Gemini 2.5 AI against indirect prompt injection attacks, enhancing its security and resilience.

Google unveiled Gemini AI upgrade, aiming to create a universal assistant with advanced planning, simulation and task management capabilities.

Gemma 3n, a new AI model designed for mobile and on-device use, offers faster, private, real-time multimodal performance, now available for developer preview.

OpenAI's deep research, powered by the o3 model, independently synthesises research, performing in minutes tasks that take humans hours.

Enkrypt AI's report reveals severe vulnerabilities in multimodal AI models, warning of safety risks from harmful prompt injections in images.

Arctic Wolf launches Cipher, an AI security assistant developed with Anthropic, offering instant insights via its Aurora Platform processing 8 trillion weekly events.

Enterprise AI use has surged over 3,000%, with ChatGPT dominating but also being the most-blocked app due to security concerns, says Zscaler report.

Tenable Research has raised alarms over security vulnerabilities in the generative AI model DeepSeek R1, warning it could simplify malware creation.

Infosys has unveiled its open-source Responsible AI Toolkit to help enterprises navigate ethical AI challenges, reinforcing the company's commitment to safe AI practices.

Tenable warns that the open-source AI model DeepSeek poses significant data exposure risks, echoing concerns over its potential misuse by cybercriminals.

Research by Enkrypt AI reveals the DeepSeek-R1 AI model poses significant security and ethical risks, being more biased and toxic than leading competitors.

Amazon Web Services has launched its AWS Trainium2-powered EC2 instances, offering improved performance for training large AI models at lower costs.