All articles tagged with #ai governance
AI agents are exploiting long-standing internet security gaps at unprecedented speed and scale, prompting new debates over liability and governance. While some argue these are not 'rogue' acts but failures of human oversight, others demand strict legal accountability for AI developers.
The newly launched America.gov AI portal, powered by Google Gemini and xAI Grok, is delivering inconsistent answers to politically sensitive questions. While it initially contradicted President Trump’s claims regarding the 2020 election and Jan. 6, it later refused to answer similar queries, citing a lack of political commentary. Experts warn that these fluctuations may stem from probabilistic model behaviors or hidden ideological guardrails, raising concerns about the reliability of a system intended to serve as the primary digital interface for federal services.
AI agents are transforming knowledge work from human-led creation to human-managed automation, but this shift risks eroding competitive differentiation and cognitive skills. While enterprises scale agent fleets, governments and firms face urgent challenges in governance, security, and maintaining human oversight.
Yale and Oxford researchers propose a 1,000-person global citizens' assembly to set AI safety limits, arguing that current CEO pledges and national rivalries fail to ensure coordination. The Verge highlights the technical trade-offs of isolating AI systems, noting that while air-gapping reduces risk, it compromises realistic testing and may not prevent all breaches.
At the UN General Assembly, UK Labour figure Andy Burnham met Donald Trump to urge de-escalation of US-Iran tensions, back stronger Ukrainian air defenses, and announce limited UK military support for Saudi Arabia amid Yemen attacks. Burnham also urged a global framework for artificial intelligence, arguing Britain could be an honest broker between the US and Europe; Trump rejected such a 'globalist' AI regime. Burnham signaled the UK may take further steps to reset Brexit-era discord with Europe as the next election approaches.
Twenty countries and the European Union urged international cooperation to keep AI under human oversight, proposing a possible global body to set and enforce standards, share safety incident reports, and convene states when capabilities exceed thresholds. The US and China did not sign, highlighting a divide as AI safety debates intensify ahead of the UN General Assembly.
As the United Nations opens its annual General Assembly, officials describe a world in turmoil with shifting friendships and rising stakes, focusing on issues from maritime security in the Strait of Hormuz to governance of rapid AI advances; leaders such as Donald Trump, Volodymyr Zelenskyy and Iran’s Masoud Pezeshkian are expected to hold sideline talks to break deadlocks and restore predictability.
At the United Nations General Assembly, Trump returns to the world stage amid growing anxiety about Iran's trajectory, the governance of artificial intelligence, and the future of the UN as a forum for diplomacy, against a backdrop of Middle East energy tensions and regional flashpoints. The piece notes oil-market jitters from attacks and blockades, Houthi activity threatening Red Sea shipping, and ongoing diplomacy around Greenland and Russia–Ukraine, underscoring a high-stakes effort to shape energy security, sanctions, and global security before a slate of meetings with allies and rivals.
UN Secretary-General Antonio Guterres says the UN has limited leverage to broker peace in Ukraine and the Gaza conflict, but remains committed to defending the UN Charter and international law, pursuing multilateral cooperation and guardrails for AI, and pressing for reforms to keep the organization funded. He plans to raise Ukraine and Middle East peace efforts with leaders, including a meeting with President Trump at the UN General Assembly, as the global body navigates financial strain and seeks a renewed, values-based role on the world stage.
As Trump and Xi prepare for a summit, analysts say their talks could define global AI safety standards amid U.S.–China rivalry, with Beijing pushing new governance blocs (BRICS, World AI Cooperation Organization) and Washington seeking international guardrails, while warning against containment and pursuing the edge in AI capabilities.
OpenAI CEO Sam Altman told a Salesforce conference that while fear of AI's rapid progress is justified, people should trust the industry to act responsibly and push for safety, with some leaders arguing for self-regulation rather than government rules. The discussion mirrors broader industry debates, as figures like Bernie Sanders and Steve Bannon call for governance, Nvidia’s Jensen Huang and Meta’s Mark Zuckerberg push for internal alignment, and Anthropic/OpenAI signal ongoing voluntary safety standards with other labs amid political friction.
Anthropic CEO Dario Amodei proposes creating Model Evaluation and Threat Research (METR), a third-party watchdog that would have employee-like access to oversee safety practices, model training, and incident reporting across AI labs. METR traces its roots to the Effective Altruism movement and is funded by EA-aligned donors, raising concerns about independence and conflicts of interest from critics who question whether a group tied to AE figures and investors can be truly impartial. Amodei argues the measure would help regulate progress and improve transparency, while regulators and industry peers weigh in with mixed reactions as debates over AI safety governance continue.
Donald Trump dismissed the need for AI guardrails, saying a strong, smart president can keep AI in check and signaling possible penalties for AI companies; his stance comes as industry leaders call for caution and regulation, and polls show broad public concern about AI and its impact on communities and data centers.
Microsoft issued a provisional code of conduct for its AI models to limit capabilities, prioritize human oversight, and seek outside input before an update slated to begin development in 2027; the move mirrors calls by industry leaders to slow AI progress, and MAI models must follow user objectives, avoid dangerous or explicit content, and not reveal their internal reasoning, with guidelines crafted after consulting experts in law, ethics, linguistics and philosophy.
At a BRICS summit in New Delhi, Xi Jinping proposed an open-source AI zone for BRICS led by the bloc, aiming to boost collaboration on large-language models, fund specialized AI research and training, and establish a broad, consensus-based global AI governance framework to foster an open AI ecosystem.