Weekly Review 14 February 2025

Some interesting links that I Tweeted about in the last week (I also post these on Mastodon, Threads, Newsmast, and Bluesky):

1. More proposed legislation to protect kids from AI chatbots: https://www.theverge.com/news/605728/california-chatbot-bill-child-safety
2. Meta says it will not release a dangerous AI. But Meta is the one deciding what is and isn't dangerous: https://www.computerworld.com/article/3816687/meta-promises-not-to-release-dangerous-ai-systems.html
3. Adopting AI can require a significant outlay on new and specialised hardware: https://www.informationweek.com/it-infrastructure/the-cost-of-ai-infrastructure-new-gear-for-ai-liftoff-
4. Even though investors are skeptical about AI, they are still putting money into the hardware to run it: https://www.nextplatform.com/2025/02/05/skepticism-about-ai-use-does-not-yet-negate-the-appetite-for-ai-hardware/
5. OpenAI and Meta probably are over-valued in terms of their AI offerings, but I don't think DeepSeek alone is going to knock them out of dominance just yet: https://futurism.com/openai-low-key-panicking-deepseek
6. Companies should be forced to report their energy and water usages, as the energy and cooling demands of AI continue to increase: https://www.theguardian.com/technology/2025/feb/07/call-to-make-tech-firms-report-data-centre-energy-use-as-ai-booms
7. The hidden costs of securing AI: https://www.informationweek.com/cyber-resilience/the-cost-of-ai-security
8. While this AI might be able to recognise if a PDF is a legal contract, would you trust it to interpret it for you? https://www.theverge.com/news/605741/adobe-acrobat-ai-assistant-contract-features-availability
9. While AI is expected to take up to 20% of IT budgets this year, there is still uncertainty around the return on that investment: https://www.theregister.com/2025/02/06/lenovo_ai_report/
10. How generative AI is being used to compromise security: https://www.informationweek.com/cyber-resilience/how-are-threat-actors-using-adversarial-genai-
11. DeepSeek has the worst rating in tests on AI safety: https://techcrunch.com/2025/02/07/anthropic-ceo-says-deepseek-was-the-worst-on-a-critical-bioweapons-data-safety-test/
12. Another company cuts staff because of AI: https://www.theregister.com/2025/02/05/workday_restructure_job_cuts/
13. Google plans to spend $75 billion on AI infrastructure this year. How much of this is "if you build it they will come" thinking? https://www.theverge.com/news/606242/google-q4-2024-earnings
14. Google has abandoned its "no AI in weapons" position: https://www.theverge.com/news/606418/google-ai-principles-weapons-surveillance
15. An AI that just wastes phone scammers' time, keeping them busy and stopping them scamming real people: https://www.theguardian.com/money/2025/feb/04/ai-granny-scammers-phone-fraud
16. The emerging legal issues around AI: https://www.informationweek.com/cyber-resilience/what-types-of-legal-liabilities-are-emerging-from-ai-
17. A new physics-inspired AI model: https://dataconomy.com/2025/02/07/how-physics-inspired-ai-is-making-our-roads-safer/
18. The shortage of talent in AI is driving companies to cut costs elsewhere to afford the salaries top candidates demand: https://www.informationweek.com/it-leadership/the-cost-of-ai-talent-who-s-hurting-in-the-search-for-ai-stars-
19. An AI that can generate proofs in geometry: https://techcrunch.com/2025/02/07/deepmind-claims-its-ai-performs-better-than-international-mathematical-olympiad-gold-medalists/
20. The energy demands of AI are growing at such a rate that it poses a significant risk to security of energy supply: https://www.theregister.com/2025/02/07/datacenter_energy_goldman_sachs/
21. While AI can be used to generate music, it can also be used to restore old performances that would otherwise never see the light of day: https://www.theverge.com/news/604970/beatles-ai-restored-song-now-and-then-grammy-win
22. Auction house Christie's is planning to have an auction of only AI-generated art. Artists are not happy: https://techcrunch.com/2025/02/08/christies-announces-ai-art-auction-and-not-everyone-is-pleased/
23. I suspect many Western governments will consider Chinese AI too risky to have on their systems: https://www.theregister.com/2025/02/05/australia_deepseek_ban/
24. Like it should be with all things, governmental policy on AI should be based on reality not fiction: https://techcrunch.com/2025/02/08/ai-pioneer-fei-fei-li-says-ai-policy-must-be-based-on-science-not-science-fiction/