Plus: reports of AI escaping users’ control nearly doubled in a month.
This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology.
Fertilizer is crucial for the global food supply, but making it uses a lot of energy and produces a lot of emissions. Some companies hope microbes can help, by providing crucial nitrogen to help plants grow, potentially reducing the need for chemical fertilizer.
Startup Switch Bioworks is taking a new approach: using a genetic switch that allows microbes to establish themselves and grow before shifting into nitrogen-producing mode.
The company is currently trialing its product in six US states. And, although it’s too early to say exactly how well the approach works in the field, its modeling suggests microbes could eventually replace about 50% of synthetic fertilizer.
Read the full story on plans to clean up agriculture with engineered microbes.
OpenAI has released a postmortem technical report on last month’s Hugging Face hack. But it’s missing a critical dimension: the role company culture may have played.
That’s all the more concerning because the report’s few references to human error suggest OpenAI’s culture played a big role. Its employees noticed models communicating with one another during training and evaluation, yet allowed them to continue. At multiple points, they either failed to raise the alarm or were not heard when they did.
“All these different failures are all pointing in the same direction, which is that the safety culture at OpenAI doesn’t exist or is anemically weak,” says Zvi Mowshowitz, a popular AI safety writer.
Find out what the Hugging Face hack could show about OpenAI’s culture.
This story is from The Algorithm, our weekly AI newsletter. Sign up to receive it in your inbox every Monday.
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 Reports of AI escaping users’ control nearly doubled in a monthMore than 300 cases were recorded in July, almost twice June’s total. (Guardian)+ Anthropic says it paused some AI training after Claude went rogue. (Axios)+ The media may be underplaying the risks of rogue AI agents. (Information $)+ Here’s why OpenAI agents hacked Hugging Face. (MIT Technology Review)
2 The FTC and 22 state AGs say Amazon secretly inflated ad pricesThey’ve sued the company for allegedly misleading advertisers. (Verge)+ They claim Amazon changed auctions after advertisers had bid. (CNBC)+ And say the practice cost advertisers more than $20 billion. (WSJ $)
3 Sony and Warner have sued Anthropic over songs used to train AIThey accuse the company of “blatant theft” in a major new lawsuit. (Reuters $)+ And claim that Anthropic pirated thousands of copyrighted songs. (Axios)+ AI-generated music is getting hard to spot. (MIT Technology Review)
Source: Read the original article on www.technologyreview.com
