Weekly Review 25 July 2025

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

1. The "fair use" defence of AI companies scraping the web for training data is not valid in the EU: https://www.theregister.com/2025/07/14/eu_genai_fair_use/
2. For safety, we need to know what AI are thinking: https://techcrunch.com/2025/07/15/research-leaders-urge-tech-industry-to-monitor-ais-thoughts/
3. Again, this isn't authors using prompts to trick people, it's authors tricking the AI reviewers are using instead of doing the reviewing themselves: https://www.techrepublic.com/article/news-hidden-ai-prompts-academic-research-papers/
4. For some developers, AI coding tools make them less productive, not more so: https://techcrunch.com/2025/07/11/ai-coding-tools-may-not-speed-up-every-developer-study-shows/
5. Another security bug-since fixed-in an AI: https://techcrunch.com/2025/07/15/meta-fixes-bug-that-could-leak-users-ai-prompts-and-generated-content/
6. I still use the terminal for a lot of programming work. I'm strangely gratified to see AI making its way there: https://techcrunch.com/2025/07/15/ai-coding-tools-are-shifting-to-a-surprising-place-the-terminal/
7. Energy demands of AI are rising, but so are solutions: https://www.nzherald.co.nz/world/companies-focus-on-ways-of-achieving-energy-efficiency-as-consumption-keeps-increasing/TFNBHB2PMRH5ZM726GYS4OUHV4/
8. The corporate risks around AI are beginning to be better known: https://www.theregister.com/2025/07/15/sec_risk_factors_ai/
9. Do not rely on AI for therapy: https://arstechnica.com/ai/2025/07/ai-therapy-bots-fuel-delusions-and-give-dangerous-advice-stanford-study-finds/
10. AI companies are under-estimating and under-preparing for the dangers of advanced AI: https://www.theguardian.com/technology/2025/jul/17/ai-firms-unprepared-for-dangers-of-building-human-level-systems-report-warns
11. An improved approach to Copilot AI viewing your screen, but I'm not convinced the security and privacy issues have been adequately addressed: https://dataconomy.com/2025/07/16/copilot-vision-now-sees-your-whole-screen/
12. If you're in AI, big corporations are the best place to find big salaries: https://www.theregister.com/2025/07/15/janco_it_salary_report/
13. AI chatbots are not an adequate or safe alternative to human therapists. They are a lot cheaper though: https://techcrunch.com/2025/07/13/study-warns-of-significant-risks-in-using-ai-therapy-chatbots/
14. Big AI companies are as vulnerable to the risks of growing too quickly as any other organisation: https://techcrunch.com/2025/07/15/a-former-openai-engineer-describes-what-its-really-like-to-work-there/
15. The European Union's rules around AI: https://www.techrepublic.com/article/news-eu-ai-code/
16. An AI generated band found success, and people never even realised it was AI: https://www.theguardian.com/technology/2025/jul/14/an-ai-generated-band-got-1m-plays-on-spotify-now-music-insiders-say-listeners-should-be-warned
17. AI generated an improved recipe for low-carbon concrete: https://www.theregister.com/2025/07/17/meta_ai_concrete_model/
18. AI generated slop is being used to populate click-harvesting websites: https://www.rnz.co.nz/news/in-depth/567221/how-a-local-website-was-hijacked-and-filled-with-ai-generated-coherent-gibberish
19. Ways AI can be applied to financial forecasting: https://www.datasciencecentral.com/how-ai-is-transforming-financial-forecasting-for-smbs-and-enterprises/
20. Turns out, the problems with using AI in sports are more to do with the people using it: https://www.theguardian.com/sport/2025/jul/15/rise-of-the-machines-ai-outrage-technology-tennis-sport
21. Improved algorithms boost the performance of AI: https://www.theregister.com/2025/07/17/new_algorithms_boost_ai_perf/
22. AI is making things worse for new graduates hunting for jobs: https://www.theguardian.com/technology/2025/jul/13/workforce-crisis-key-takeaways-graduates-ai-jobs-market
23. Ten simple maths and stats commands in Python: https://www.kdnuggets.com/10-python-math-statistical-analysis-one-liners
24. Why am I not surprised that this particular AI company has lax safety standards? https://techcrunch.com/2025/07/16/openai-and-anthropic-researchers-decry-reckless-safety-culture-at-elon-musks-xai/