Greetings all, and happy Wednesday.
I hope you’re all having a wonderful week. Its a beautiful blue sky day in East London. It’s almost enough to distract from whats going on in the rest of the world. Almost.
The Smithsonian has an interesting piece on how AI can help battle wildfires. Most of the information won’t be new to any of ye, but i was surprised to learn that there is a network of 1,400 ultra-high definition cameras in california, which are used to analyze objects, identify behaviour patterns, and identify faces. They can see up to 60 miles on a clear day, and have 360 degree capability. Quite the surveillance infrastructure.
The FT has an interesting piece on how Anthropic developed a ‘jailbreak’ to stop their AI models revealing dangerous behaviour, like how to develop chemical weapons. This seems like a good thing. On the health and human rights front, the Guardian also has a piece on the NHS launching the world’s biggest trial of AI breast cancer diagnosis. This also seems like a good thing.
My colleague at Queen Mary, Elke Schwarz, has a piece in Just Security on the UK’s AI Opportunities Plan, which is a nice primer/critique. As she notes, it will be really interesting to see what the plan means for (human rights) impact assessments (and of course, compliance with the Public Sector Equality Duty) and how that might play out in the courts in a few years time.
The big regulatory news of the week, of course, is that AI systems classified as ‘unacceptable risk’ are now formally prohibited under the EU AI Act. The European Commission have published their Guidelines on Prohibited AI Practices, which i have yet to read but am sure is on all of our agendas.
Thats me for the week. Well, newsletter wise at least, the work week – unfortunately – has a long way left to run.
Thanks, as ever, to Ameneh Deshiri.
I’ll leave you with this Late Night Tales mix by Jon Hopkins. Its beautiful, and calm, i’ve been listening to it as i’ve put this post together.
Stay well.
— — —
Asia Times, How DeepSeek revolutionized AI’s cost calculus – Asia Times
The Guardian, NHS to launch world’s biggest trial of AI breast cancer diagnosis
Financial Times, Anthropic makes ‘jailbreak’ advance to stop AI models producing harmful results
The New York Times, OpenAI Unveils A.I. Tool That Can Do Research Online
WIRED, Here’s How DeepSeek Censorship Actually Works—and How to Get Around It
WIRED, DeepSeek Has Gotten OpenAI Fired Up
The Register, TSA facial recognition technology faces audit probe • The Register
Rest of World, Meet Eva, the AI chatbot based on a woman in prison
Algorithm Watch, A Dual Track Approach to Systemic Risks ensures flexibility and transparency in the Digital Services Act
BBC, Google owner Alphabet lifts ban on its AI being used for weapons
Smithsonian, How A.I. Can Help Humans Battle Wildfires, From Advanced Camera Systems to Forecasting Models | Smithsonian
DefenseOne, How DeepSeek changed the future of AI—and what that means for national security
Reuters, Italy’s regulator blocks Chinese AI app DeepSeek on data protection
Reuters, French privacy watchdog to quiz DeepSeek on AI, data protection
Euronews, EU lacks guidance on AI prohibitions, as rules start to apply
The Register, AI facial recognition could sink this murder probe
The Conversation, AI is bad for the environment, and the problem is bigger than energy consumption
The Conversation, From chatbot to sexbot: What lawmakers can learn from South Korea’s AI hate-speech disaster
The Guardian, Apple to report earnings for first quarter of 2025
The Guardian, AI tools used for child sexual abuse images targeted in Home Office crackdown
The New York Times, DeepSeek’s A.I. Chatbot Awkwardly Navigates China’s Censors
The Guardian, US Authors Guild to certify books from ‘human intellect’ rather than AI
Hurriyet Daily News, Türkiye develops AI system to curb repeat offences
Euronews, Belgium to harness AI to find criminals with first dedicated minister
Financial Times, What to do when the computer says no
EDRi, Why Ireland is the Achilles Heel of the EU’s Fightback against Big Tech
Rest of World, Is Google Maps fatally misleading drivers in India? It’s complicated
Irish Times, What data is your car storing about you?
AI and Human Rights Policy News
The Free Press, New Vatican document offers AI guidelines from warfare to health care
European Commission, Updated Frequently Asked Questions about the Data Act
Open Rights Group, Safety Not Surveillance Coalition (I missed this previously)
Blogs
Just Security, What DeepSeek Really Changes About AI Competition
Just Security, Throwing Caution to the Wind: Unpacking the U.K. AI Opportunities Action Plan
Google, Responsible AI: Our 2024 report and ongoing work (the AI weapons edition)
Privado, DeepSeek App: A Closer Look at Its Privacy Posture
Lawfare, The Trouble With AI Safety Treaties
SoftSkills, China’s Social Credit System: Workplace Implications of AI Surveillance
The Mac Observer, Group Asks Apple to Explain Where It Gets AI Data and How It Uses It
JD Supra, Looking Ahead to 2025 Data Privacy Trends and Developments
Forbes, The ‘Doomsday Clock’ Ticks Closer To Global Catastrophe—And AI Is Now A Primary Concern
Podcast:
Just Security, The Just Security Podcast: Diving Deeper into DeepSeek
Video:
TeachPrivacy, Webinar – AI and Privacy Collide
Reports
DSIT of UK, International AI Safety Report
Thomas Reuters Foundation, Journalism in the AI Era: A TRF Insights surveyNew America, Exploring the Intersection of OSINT and Data Privacy in the Digital World