On February 12, 2026, Maryland legislators proposed the Maryland Artificial Intelligence Toy Safety Act, aimed at regulating AI-enabled toys to ensure child safety. The act would impose pre-market compliance obligations and data privacy requirements, alongside mechanisms to protect children from harmful content and marketing practices.
In his remarks to the Westminster Legal Policy Forum, Sir Brian Leveson emphasized that while AI can help address delays in the criminal courts, it must be implemented with strict safeguards to support, not replace, human decision-making. He warned against the risks of fragmented systems and stressed the importance of coordinated governance and AI literacy across the legal framework.
The New York State Senate is advancing a bill that would hold AI chatbot proprietors liable for offering professional advice that constitutes unauthorized practice. This legislation aims to ensure that only licensed human professionals provide substantive responses, while also facilitating civil lawsuits against chatbot owners for violations.
Prompt injection attacks exploit the vulnerabilities of AI language models by tricking them into performing tasks they are not designed to do, similar to traditional hacking methods. These attacks reveal the lack of human-like judgment and context in AI models, making them susceptible to manipulation through cleverly crafted prompts.
The Trump administration is considering new regulations that would require foreign companies importing American AI chips to invest in the U.S. and acquire prior authorization for exports. This move aims to strengthen control over AI chips and leverage them in diplomatic negotiations.
TransferMate has launched Vivox AI agents globally to automate its Know Your Business (KYB) compliance processes, facilitating rapid expansion while maintaining regulatory standards. This innovative deployment enhances efficiency and oversight in compliance operations, reflecting a shift towards responsible AI in financial services.
As India integrates AI into justice and welfare, the focus has shifted toward “Responsible AI,” governed by the IndiaAI Governance Guidelines. This approach emphasizes transparency and ethics to prevent algorithmic bias, ensuring that digital tools do not inadvertently exclude marginalized communities.
In a high-stakes confrontation between AI ethics and national security, Anthropic has been excluded from Pentagon contracts after refusing to relax its ethical safeguards. As tensions rise, OpenAI steps in with its own agreement, emphasizing safety measures while raising questions about the boundaries of corporate governance and national security.
In February 2026, the Spanish data protection authority (AEPD) issued guidance on the data protection implications of using AI agents, clarifying the responsibilities of controllers and processors under GDPR. The guidance emphasizes that while AI agents can autonomously handle data, the legal accountability for processing remains with the organizations deploying these systems.
Center-right and center-left EU lawmakers are negotiating a potential deal on the AI Act that would relax rules for industrial applications in exchange for a ban on sexual deepfakes. This proposal is set to face opposition from the Greens and uncertainty regarding negotiations with EU governments ahead of a committee vote on March 18.