North Dakota is turning its cold climate into a testing ground for heat recycling innovation. The Legendary Harvest Project, a new collaboration between North Dakota State University (NDSU) and the Research Innovation Institute, explores whether waste heat from data centres could be used to support year-round food production in harsh northern conditions. Modern data centres tend to emit large volumes of low-grade heat (usually 25-35°C), which the NDSU team plans to redirect into adjacent greenhouses to create a closed-loop system where digital infrastructure heats physical production.
At its core, the project is a real-world example of what we call “industrial symbiosis”: the process of turning an industry’s byproduct into inputs for another industry. Europe seems to be pioneering this concept. For example, in Rjukan, Norway, Green Mountain supplies excess heat from its data center to warm water for a land-based trout farm. Similar concepts are being explored in France, where a proposed data centre near Paris would heat greenhouses and aquaculture facilities. In cold regions like North Dakota, where harsh winters and short growing seasons limit food production, this model could offset the burden of a large data centre footprint by delivering carbon reductions and improved local food security.
It could also offer an example of how data centres are located and designed differently. Instead of being isolated energy consumers, the future facilities could be designed as key parts of larger industrial hubs.
AI agents are no longer just answering questions: they are now browsing the web, filling forms and, in some cases, are executing transactions on the user's behalf. But according to the latest update of the MIT’s AI Agent Index, basic safety disclosure is “dangerously lagging” for the systems with the highest autonomy.
Out of the 30 “state-of-the-art” chat, browser and workflow agents studied, only four have released detailed documentation specifying autonomy levels, behavioural constraints, and real-world risks. Only five companies provide data results proving they checked how safe the bot is, and only seven provide proof of third-party testing – exactly the kind of empirical evidence regulators would likely assess. Browser-based agents – which can click, purchase tickets, or submit forms – show the highest rate of missing safety disclosures, according to the study.
More troubling still, the report states that at least six agents in the Index deliberately mimic human browsing patterns and bypass anti‑bot protections. Most of them do not identify themselves as AI to websites or end users. In practice, this means that websites would increasingly be unable to distinguish human visitors from legitimate agents or aggressive scrapers.
The implication is clear: the agent ecosystem is rapidly evolving, but governance has not kept pace. A 2026 survey from Drexel University found that 41% of organisations are already using agentic AI in daily operations, yet only 27% report having governance frameworks mature enough to monitor and manage these systems effectively. Combined with the results from the MIT’s AI Agent Index, this suggests that there’s an overall mismatch: AI agents gain real decision power faster than organisations and regulators can set rules, monitor their actions, and put effective safeguards in place.
A recent interactive report by Reuters reveals how modern scam operations are starting to look less like random or isolated fraud and more like structured tech workflows at scale. Handbooks found during law enforcement operations in the Philippines describe repeatable processes meant to identify target victims, build fake relationships with them, and guide them toward fraudulent investment platforms. Scammers exploit loneliness and social isolation – especially among seniors and emotionally vulnerable adults – using AI-driven chatbots, translation tools, and synthetic personas to mimic real online interactions and manage dozens of conversations simultaneously. The result is “scalable,” personalised scams, as generative AI lowers the barrier to social engineering.
Source: A scammer's blueprint - Reuters - Illustration by Ally J. Levine
If you like Tech+10, please share it with your friends. Have a suggestion? Let us know.
