Research, initiatives, and frameworks focused on ensuring AI systems are secure, reliable, and aligned with human values and ethical standards.
FlexRule has launched its Decision Governance platform to help enterprises make decisions explicit, owned, and explainable across manual, automated, and AI operations. The platform aims to close long-standing governance gaps in organizational decision-making.
SCRT Labs has integrated Intel Trust Authority into every SecretVM, enabling default attestation and independent verification for confidential computing workloads.
Pearl Enterprise evaluated 25 AI models from OpenAI, Anthropic, Google DeepMind, Microsoft, and others, finding that even top systems align with expert judgment only about 70% of the time, with performance dropping to as low as 20% in some domains.
EquiLend has acquired Finadium, a research and consultancy firm in securities finance and capital markets. Finadium will operate as an independent subsidiary under current leadership, maintaining editorial independence while expanding its services.
Common Sense Media has launched the Youth AI Safety Institute, an independent organization that will test AI products used by children, publish the results, and establish safety standards to ensure they are developmentally appropriate.
OpenBox AI and Mastra have partnered to integrate runtime governance directly into the Mastra TypeScript agent framework, enabling compliance-ready oversight and audit capabilities by default.
CollectivIQ has announced major upgrades to its AI consensus platform, adding multimodal image generation, integrated payment capture, and expanded retrieval capabilities to enhance accuracy, collaboration, and governance.
OPAQUE has acquired cryptographic AI technologies from Abu Dhabi’s Technology Innovation Institute, expanding its Confidential AI platform with post-quantum protections for secure model training and deployment.
ITRAK 365 has launched ITRAK IQ, an AI integration framework designed to connect Environment, Health and Safety (EHS) and operational data with AI assistants within a customer's Microsoft environment. The platform enables secure, governed AI use without sending data to external services.
At the World Economic Forum in Davos, SandboxAQ CEO Jack Hidary warned that GPS interference and AI-enabled cyberattacks are merging into a single threat to critical infrastructure. He highlighted the company's technologies AQNav and AQtive Guard as tools to address these risks.
A new report by the National Safety Council and Wolters Kluwer Enablon shows that AI use in environmental, health, and safety programs is expanding quickly, though most professionals call for clear guardrails and stronger digital foundations.
KBR has secured a $200 million contract from the U.S. Department of Transportation's Volpe Center to enhance aviation safety and modernize transportation systems using AI and data analytics.
NSS Labs has released the AI Protection Systems test methodology, a framework for evaluating the security of enterprise AI deployments across eight key dimensions, including prompt injection resistance and data exfiltration prevention.
Matt O'Brien, CEO of Snow Crash Labs, said in a podcast interview that the U.S. is falling behind China in power capacity needed for AI infrastructure and that enterprises are deploying models without adequate quality control, creating major safety and compliance risks.
Singapore has proposed ISO/IEC 42119-8, the first international standard for testing generative AI systems. The standard focuses on benchmarking and red teaming and will be discussed at the ISO/IEC JTC 1/SC 42 plenary meeting in Singapore.
Kroll reports that three quarters of organizations have experienced security incidents linked to AI applications or models in the past two years, with over a quarter incurring costs exceeding one million dollars.
Inflectra has launched a free Early Access Program for SureWire, a new quality assurance platform built to test non-deterministic AI agents. The platform introduces specialized testing and judge agents to identify vulnerabilities in AI systems before deployment.
Google has introduced new mental health safety features in its Gemini AI system and pledged $30 million to support global crisis helplines. The updates include a redesigned help interface, expanded partnerships, and safeguards for younger users.
The PyTorch Foundation has added Safetensors, developed by Hugging Face, as its newest project to improve secure AI model distribution and prevent arbitrary code execution risks.
Chai AI has deployed a 5,000+ GPU cluster to enhance post-training research focused on large-scale model alignment and safety. The company also formalized new global compliance measures, including age gating and regional verification protocols.
Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.
