NVIDIA's New Robotics Safety System, US Approves Anthropic's Mythos 5, and AI Cyber Risks Warning

Hi there,

In this edition of the AI Policy Brief, we bring you a range of developments in the field of AI regulations and safety standards. Grow Therapy and Stanford University have teamed up to create guidelines for the safe use of AI in mental health care, aiming to reduce potential harm during crisis situations. Meanwhile, the AI Alliance and BharatGen have introduced Project Tapestry, which focuses on advancing AI capabilities while ensuring that countries maintain control over their data and AI models.

Additionally, L8P8, Inc. has unveiled Loop8ID Age Shield, an AI-driven age verification platform designed to comply with U.S. child safety laws without collecting personal data. Darkstrike has brought on AI safety expert Roman Yampolskiy as an advisor to bolster its AI governance strategies. In legislative news, North Dakota has formed a committee to study AI and data center regulations ahead of the 2027 legislative session. Lastly, VIDIZMO has launched an AI Intelligence Hub to help organizations manage AI models in compliance with regulatory standards like CJIS and FedRAMP.

FIU Researchers Find AI Image Processing Flaws

Researchers from Florida International University have found that small changes to images can deceive AI systems, leading to potentially harmful outputs. This discovery underscores the need for enhanced security in AI applications. Read more

FORT Robotics Partners with NVIDIA for Robot Safety

FORT Robotics has partnered with NVIDIA to enhance safety in autonomous robots by joining the NVIDIA Halos for Robotics ecosystem. They will present a new safety application at the Automate conference in Chicago. Read more

Five Eyes Warns of AI Cyber Risks

The Five Eyes intelligence alliance has issued a warning about the cyber risks posed by advanced AI models, emphasizing their potential to enhance hacking capabilities and urging immediate action to strengthen defenses. Read more

Grow Therapy and Stanford Partner on AI Safety in Mental Health

Grow Therapy and Stanford University have partnered to develop standards for safe AI use in mental health care, focusing on minimizing harm and ensuring patient safety. Read more

AI Alliance and BharatGen Launch Project Tapestry

The AI Alliance and BharatGen have initiated Project Tapestry to advance AI capabilities while ensuring national control over data and models. This was discussed at the G7 Summit to promote global AI sovereignty. Read more

ASML Faces Challenges with US MATCH Act

Dutch Trade Minister Sjoerd Sjoerdsma has expressed concerns in Washington about the US MATCH Act, which could restrict Chinese access to semiconductor equipment and impact ASML's sales to China. Read more

Darkstrike Appoints AI Safety Expert as Advisor

Darkstrike has announced the appointment of AI safety expert Roman Yampolskiy as an advisor to enhance their governance strategies. Read more

Legion LegalTech Corp Sues US Government Over AI Model Access Restrictions

Legal technology company Legion LegalTech Corp has filed a lawsuit against the US government, challenging a directive that limits foreign access to Anthropic's AI models, Fable 5 and Mythos 5. Read more

North Dakota Forms AI and Data Center Study Committee

Lawmakers in North Dakota have established an interim committee to study artificial intelligence and data center regulations in preparation for the 2027 legislative session. The committee will explore regulations, child safety measures, and the impact of data centers on local resources. Read more

Cyphlens Appoints Jon Finer and Launches CypherFile

Cyphlens has appointed Jon Finer to its advisory board and introduced CypherFile, a new AI-driven encryption product aimed at improving data security. Read more

PersonaShield Launches Platform for Likeness Control

PersonaShield, founded by former Apple executive Phillip Shoemaker, has introduced a platform for creators to manage their likeness in AI-generated content. The platform provides tools for monitoring unauthorized use, setting usage guidelines, and monetizing authorized content. Read more

Trump Administration Grants Access to Anthropic Mythos 5

The Trump Administration has lifted restrictions on the Anthropic Mythos 5 model, allowing over 100 US companies and agencies, including non-American employees, to access it. This follows a previous ban due to security concerns. Read more

BIS Warns of AI Investment Risks

The Bank for International Settlements has warned that the rapid increase in AI investments could pose global economic risks, likening it to historical economic bubbles. Read more

Austria Urges EU to Host Anthropic PBC

The Austrian government is encouraging the EU to host Anthropic PBC to address US restrictions on foreign AI model access. Read more

MeetKai Brasil Secures SERPRO AI Contract

MeetKai Brasil has been chosen by SERPRO to spearhead Brazil's Sovereign AI initiative, focusing on developing a secure national AI infrastructure. Read more

VIDIZMO Launches AI Hub for Regulatory Compliance

VIDIZMO has launched its AI Intelligence Hub, enabling organizations to control model origin and data location on their own infrastructure. This addresses regulatory challenges with foreign-origin models amid increasing government restrictions. Read more

Ohio Requires AI for Medicaid Compliance

On June 10, 2026, Ohio passed Senate Bill 315, mandating the use of automated AI solutions by the Ohio Department of Medicaid to enhance compliance and reduce improper payments. Read more

California and Anthropic Sign AI Tool Agreement

The California Government has partnered with Anthropic to offer its AI tool, Claude, to state agencies at a 50% discount, including free training and support, to improve government services. Read more

Nvidia Launches Halos for Robotics Safety System

Nvidia has unveiled Halos for Robotics, a safety system aimed at improving AI compute and safety in robotics. Agility is the first to use it in their humanoid robots for industrial purposes. Read more

Garfield AI Aids HR Consultant in Winning Court Case

HR consultant Tamires Camal Taquidir won a court case in England with the help of the AI law firm Garfield AI, marking a legal first. The AI firm managed pre-trial work for her claim of unpaid fees, leading to a favorable judgment. Read more

L8P8, Inc. Launches AI Age Verification Solution

L8P8, Inc. has launched Loop8ID Age Shield, an AI-powered platform for age verification to aid compliance with U.S. child safety laws, using device signals from Apple and Google. Read more

Trump Administration Requests Limited Release of GPT-5.6

The Trump administration has asked OpenAI to initially release its GPT-5.6 AI model only to government-approved partners, citing concerns over the model's advanced capabilities. Read more

U.S. Bill Mandates AI Incident Reporting

U.S. Representative Nathaniel Moran has introduced a bill requiring AI companies to report critical incidents to the U.S. Commerce Department within a week, addressing national security and public safety concerns. Read more

US Approves Anthropic's Mythos 5 AI Model for Limited Use

Anthropic has been granted approval by the US government to allow certain trusted partners access to its Mythos 5 AI model after resolving national security issues. The decision comes after previous restrictions on foreign nationals' access. Read more

Proception Settles Lawsuit with Tesla, Secures Funding

Proception has settled a trade secret lawsuit with Tesla and announced an $11 million seed funding round led by First Round Capital. The company focuses on developing robotic hands and has started shipping its first products. Read more

Categories

Companies

Resources

NVIDIA's New Robotics Safety System, US Approves Anthropic's Mythos 5, and AI Cyber Risks Warning - AI Policy Brief #75

We hope you enjoyed this article.

More from: Regulation

More from: Data Protection & Privacy

Subscribe to AI Policy Brief