hotai.news
AI Briefings / Date Archive

2026-07-28
AI Briefings Daily Report.

An archive of 20 AI news briefs collected on 2026-07-28. Check the original sources for details and updates.

This content still needs to be cross-checked against the original publication page.

20 curated selectionsView other dates ↗
04News

US professor uses hidden prompts to catch AI cheating, 32 of 35 exam papers caught

Jason Gibson, a history professor at Alcorn State University in the US, devised a clever trick to catch AI cheating: embedding hidden prompts about 'Madagascar' in white font within an Industrial Revolution exam document. As a result, 32 of 35 exam papers contained illogical content such as 'a purple bicycle in Madagascar whispering to the ceiling,' exposing the cheating on the spot.

Source links can be verified on the original siteRead original ↗
10News

AI companies exposed for mass acquisition of old books to obtain"Clean"Training materials

To avoid low-quality 'AI junk content', multiple AI companies are mass-purchasing secondhand books through intermediaries as high-quality training data sources. Reports say the relevant procurement scale ranges from a thousand to a million books, with no topic preference and regardless of price. This destructive scanning method is efficient, but may cause precious books to disappear permanently and trigger ethical disputes over the privatization of knowledge resources.

Source links can be verified on the original siteRead original ↗
12News

Anthropic CEO Amodei Reiterates Stance on Risks of Open-Source Weight Models, Denies Ever Calling for a Total Ban

Anthropic CEO Dario Amodei is once again warning about the risks of open AI models while insisting he has never called for a ban. He argues that authoritarian states like China could overtake the US and that open models could be misused for biological or cyberattacks. Critics say he's mostly trying to protect his own business from cheaper competition.

Source links can be verified on the original siteRead original ↗
13News

Zhiyuan WITA-Omni Preview embodied large model tops DailyOmni full-modal understanding leaderboard with composite score 85.21

The self-developed embodied native full-modal large model WITA-Omni Preview by Zhiyuan topped the DailyOmni leaderboard with 85.21 points, surpassing mainstream domestic and foreign models in audio-video joint understanding and temporal reasoning capabilities. Its core lies in the innovative Thinker–Talker–Actor architecture, which elevates physical actions and expressions to first-level outputs alongside speech.

Source links can be verified on the original siteRead original ↗
15News

The professor set up"The Hidden Trap of Prompting"Caught 32/35 students cheating with AI.

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. Facepalm: Modern problems require modern solutions – according to the meme. In this case, the issue is students using AI to cheat on assignments. The solution, used to great effect by one professor, was to hide a white-colored font in the assignment prompt telling the AI to talk nonsense about an unrelated subject.

Source links can be verified on the original siteRead original ↗
17News

Alibaba Cloud Lingjun Zhenwu M890 Super Node Instance adapts to Moonshot AI's Kimi K3 large model

Alibaba Cloud announced that its Lingjun Zhenwu M890 Super Node Instance has successfully adapted to Moonshot AI's latest flagship model, Kimi K3, with 2.8 trillion parameters. Based on T-Head's training-inference integrated AI chips and high-speed interconnect architecture, the instance can restrict expert parallel communication traffic for trillion-parameter MoE models within a single high-bandwidth domain, effectively improving inference efficiency.

Source links can be verified on the original siteRead original ↗
18News

Nvidia CEO Jensen Huang: AI eliminates tasks, not jobs

In his speech, Jensen Huang emphasized the need to distinguish between "tasks" and "jobs," noting that AI automates the former but not the latter. Citing customer service, radiologists, and software engineers as examples, he pointed out that AI boosts efficiency and could actually create more jobs. This stands in sharp contrast to pessimistic forecasts of mass unemployment caused by AI.

Source links can be verified on the original siteRead original ↗