Top of the Day
Challenges in Governing AI Agents
(Noam Kolt – Lawfare – 3 March 2025) Leading AI companies have released a new type of AI system: autonomous agents that can plan and execute complex tasks in digital environments with limited human involvement. OpenAI’s Operator, Google’s Project Mariner, and Anthropic’s Computer Use Model all perform a similar function. They type, click, and scroll in a web browser to carry out a variety of online tasks, such as ordering groceries, making restaurant reservations, and booking flights. While the performance of these agents is currently unreliable, improvements are on the horizon. Scores on multiple benchmarks are steadily improving. The aspiration is to create AI agents that can undertake a broad range of personal and professional activities, serving as artificial personal assistants and virtual coworkers. – https://www.lawfaremedia.org/article/challenges-in-governing-ai-agents
Latin American Orgs Face 40% More Attacks Than Global Average
(Nate Nelson, Dark Reading – 3 March 2025) Cyber threats are accelerating faster in Latin America than anywhere else in the world. The trend has been building for at least a year now, actually. Last summer, Check Point tracked a 53% year-over-year rise in weekly cyberattacks against organizations in the region, followed at a distance by Africa (37%) and Europe (35%). Today, the cybersecurity company reports, Latin American companies suffer 2,569 attacks per week on average — nearly 40% more than the global average of 1,848. Critical industries like healthcare, communications, and governments and militaries are frequently hounded — those organizations often face around 3,000 to 4,000 attacks per week — but even ordinary citizens are feeling the heat, primarily through their financial apps and institutions. – https://www.darkreading.com/cybersecurity-analytics/latin-american-orgs-more-cyberattacks-global-average
The battle for the internet
(Mercedes Page – The Strategies – 3 March 2025) Democracies and authoritarian states are battling over the future of the internet in a little-known UN process. The United Nations is conducting a 20-year review of its World Summit on the Information Society (WSIS), a landmark series of meetings that, among other achievements, formally established today’s multistakeholder model of internet governance. This model ensures the internet remains open, global and not controlled by any single entity. – https://www.aspistrategist.org.au/the-battle-for-the-internet/
Three Key Moments from Davos for the Cyber Leadership Community
(William Dixon – RUSI – 3 March 2025) Davos 2025 spotlighted rising geopolitical tensions, AI disruption, and shifting tech leadership – posing urgent questions for the cyber community and the UK’s global strategy. – https://www.rusi.org/explore-our-research/publications/commentary/three-key-moments-davos-cyber-leadership-community
TikTok’s Teen Data Use Probed by UK Regulators
(Becky Bracken – Dark Reading – 3 March 2025) The United Kingdom’s Information Commissioner’s Office (ICO) wants TikTok, Imgur, and Reddit to open up their algorithms and prove that they are not using teenagers’ personal data to feed them content recommendations. The ICO passed a children’s code of conduct in 2021, which requires platforms to take “age assurance measures,” including tools that help estimate a child’s age, and protect them from potentially harmful content, the ICO said in its statement. “In announcing these investigations, we are making it clear to the public what action we are currently taking to ensure children’s information rights are upheld,” John Edwards, the UK’s Information Commissioner, said in a statement about the launch of the investigation. “This is a priority area, and we will provide updates about any further action we decide to take.” – https://www.darkreading.com/application-security/tiktok-teen-data-use-probed-regulators
Getting macro-ready for the AI race
(Amit Singh, Adam Triggs – East Asia Forum – 2 March 2025) As the AI race drives substantial investment and productivity gains, optimal policy conditions are necessary to ensure maximum economic benefit. Policymakers must take advantage of the AI boom by encouraging greater savings and more efficient service delivery, fostering competition for global capital, streamlining financial regulations to direct savings where needed, and prioritising workforce mobility and skill development for thriving AI industries. – https://eastasiaforum.org/2025/03/02/getting-macro-ready-for-the-ai-race/
Generative AI, Democracy and Human Rights
(David Evan Harris, Aaron Shull – Centre for International Governance Innovation – 28 February 2025) Disinformation is not new, but given how disinformation campaigns are constructed, there is almost no stage that will not be rendered more effective by the use of generative artificial intelligence (AI). Given the unsatisfactory nature of current tools to address this budding reality, disinformation, especially during elections, is set to get much, much worse. As these campaigns become more sophisticated and manipulative, the foreseeable consequence will be a further erosion of trust in institutions and a heightened disintegration of civic integrity, which in turn will jeopardize a host of human rights, including electoral rights and the right to freedom of thought. In this policy brief, David Evan Harris and Aaron Shull argue that policy makers must hold AI companies liable for the harms caused or facilitated by their products that could have been reasonably foreseen, act quickly to ban using AI to impersonate real persons or organizations, and require the use of watermarking or other provenance tools to allow people to distinguish between AI-generated and authentic content. – https://www.cigionline.org/publications/generative-ai-democracy-and-human-rights/
AI Action Summit in Paris Highlights A Shifting Policy Landscape
(Stanford HAI – 27 February 2025) Leaders from government, international organizations, and academia headed to Paris this month for the AI Action Summit, where they engaged in important discussions on how AI can prioritize public interest. Key conversations centered around providing independent and reliable AI access, developing more environmentally friendly technologies, and promoting effective global governance. The summit week included nearly 100 events worldwide from Feb. 6-11, 2025, including an international conference on AI and society and a discussion series on AI and culture. – https://hai.stanford.edu/news/ai-action-summit-in-paris-highlights-a-shifting-policy-landscape
AI+Education Summit: The Future is Already Here
(Stanford HAI -27 February 2025) Artificial intelligence can summarize text, find bugs in code, and create images. It can even record and summarize panels at a conference.* New tools and use cases are being developed as we speak, each model more powerful than the last. What does this reality mean for teachers and learners? In its third instance, the AI+Education Summit hosted by the Stanford Accelerator for Learning and the Stanford Institute for Human-Centered Artificial Intelligence (HAI) brought together researchers, educators, tech developers, and policymakers for pivotal conversations on how to shape a thriving learning ecosystem with human-centered AI technologies. The convening showcased cutting-edge, research-based applications of AI in learning and facilitated dialogue about how to ensure AI serves education ethically, responsibly, and equitably. – https://hai.stanford.edu/news/aieducation-summit-the-future-is-already-here
Security
Polish space agency investigates cyberattack on its systems
(Daryna Antoniuk – The Record – 3 March 2025) Poland’s space agency (POLSA) announced on Sunday it had suffered a cyberattack and is currently investigating the incident. In response to the attack, the agency said it disconnected its network from the internet, and as of Monday its website remained inaccessible. Poland’s digital minister, Krzysztof Gawkowski, confirmed that state cybersecurity services had detected unauthorized access to POLSA’s IT infrastructure and had secured the affected systems. Cyber specialists are now working to identify the attackers behind the breach, he added. – https://therecord.media/poland-space-cyberattack-agency-investigate
Phishers Wreak ‘Havoc,’ Disguising Attack Inside SharePoint
(Elizabeth Montalbano – Dark Reading – 3 March 2025) A complex phishing campaign is targeting Microsoft SharePoint accounts with malicious documents aimed at getting users to compromise themselves by deploying a PowerShell command. The attack is a ClickFix-style cyberattack campaign that ultimately aims to deploy the open source and powerful Havoc command-and-control (C2) framework to gain full control over the targeted system. Havoc is a C2 framework akin to Cobalt Strike, Silver, and Winos4.0 that attackers use to gain control over targeted systems. It’s both open source and available on GitHub, making it easy for threat actors to customize it and add advanced anti-detection capabilities, the researchers said. – https://www.darkreading.com/cyberattacks-data-breaches/phishers-wreak-havoc-disguising-attack-inside-sharepoint
Defense, Intelligence, and Warfare
China’s new spy drone with 310-mile radar range can track US stealth jets
(Aamir Khollam – Interesting Engineering – 3 March 2025) China has deployed its most advanced long-range surveillance drone, the WZ-9 Divine Eagle, to the South China Sea. This move significantly boosts China’s reconnaissance capabilities in the region. Satellite imagery confirmed the drone’s presence at Ledong Air Base on Hainan Island, a key strategic location. The deployment of the WZ-9 represents a major challenge to the US air superiority, as the aircraft is specifically designed to track and counter stealth technology. The deployment of the WZ-9 Divne Eagle aligns with China’s anti-access/area denial (A2/AD) strategy, aimed at enhancing control over the South China Sea. – https://interestingengineering.com/military/china-wz-9-drone-deployed
Frontiers
Holistic Evaluation of Large Language Models for Medical Applications
(Stanford HAI – 28 February 2025) Large language models (LLMs) hold immense potential for improving healthcare, supporting everything from diagnostic decision-making to patient triage. They can now ace standardized medical exams such as the United States Medical Licensing Examination (USMLE). However, evaluating clinical readiness based solely on exam performance is akin to assessing someone’s driving ability using only a written test on traffic rules, a recent study finds. While LLMs can generate sophisticated responses to healthcare questions, their real-world clinical performance remains under-examined. In fact, a recent JAMA review found that only 5% of evaluations used real patient data and the majority of studies evaluate performance on standardized medical exams. This state of affairs underscores the need for better evaluations that measure performance on real-world medical tasks, preferably using real clinical data when possible. – https://hai.stanford.edu/news/holistic-evaluation-of-large-language-models-for-medical-applications
Generative AI Tool Marks a Milestone in Biology
(Stanford HAI – 27 February 2025) Imagine being able to speed up evolution – hypothetically – to learn which genes might have a harmful or beneficial effect on human health. Imagine, further, being able to rapidly generate new genetic sequences that could help cure disease or solve environmental challenges. Now, scientists have developed a generative AI tool that can predict the form and function of proteins coded in the DNA of all domains of life, identify molecules that could be useful for bioengineering and medicine, and allow labs to run dozens of other standard experiments with a virtual query – in minutes or hours instead of years (or millennia). The open-source, all-access tool, known as Evo 2, was developed by a multi-institutional team co-led by Stanford HAI affiliate faculty Brian Hie, an assistant professor of chemical engineering and a faculty fellow in Stanford Data Science, and was partially funded by a Stanford HAI Hoffman-Yee Grant. Evo 2 was trained on a dataset that includes all known living species, including humans, plants, bacteria, amoebas, and even a few extinct species. Stanford Report talked to Hie about Evo 2’s advanced capabilities, why the scientific world is so eager to get its hands on this new tool, and how Evo 2 could reshape the biological sciences. – https://hai.stanford.edu/news/generative-ai-tool-marks-a-milestone-in-biology