Google forces AI search on California schools

Key Takeaways

• The US Center for AI Safety has halted research and now reports to the White House, sparking independence concerns
• CAISSI operates with a $10 million annual budget and has had no permanent leader since 2022
• Google is blocking K-12 students from regular search in California schools unless districts adopt its AI version
• A South Korea study found Adobe Firefly tools speed up design but reduce creative autonomy
• The five largest US hyperscalers plan to spend $800 billion on AI infrastructure this year
• Capital Economics warns the AI investment boom may become unsustainable as revenues lag spending
• A US aircraft nearly attacked a Chinese ship after an AI chatbot produced a flawed conclusion
• Only four out of 25 organizations have practiced an AI incident response scenario
• Independent game developer NasNakarus saw over a dozen AI-assisted clones of their viral game appear on Steam
• A new report compares five security products for detecting unauthorized Shadow AI use at work

White House oversight raises concerns for AI safety center

A federal AI safety office called CAISSI has stopped releasing public research reports. Officials were asked to report informally to a White House office. Critics worry this move could reduce the independence of AI safety research. The center has a $10 million annual budget and has faced leadership changes. The White House pointed to a new Super Intelligence Force as a reason for the arrangement.

US AI safety staff report to White House amid independence fears

The US Center for AI Safety has no permanent leader since 2022. Its research and publications have been paused. Staff are now reporting directly to the White House, raising concerns about political influence. The center was created in 2020 to oversee safe AI development. Experts say the lack of leadership and halted research could weaken its ability to monitor AI risks like bias and cybersecurity threats.

AI safety center reports to White House, sparking independence debate

The NIST AI Safety Center has operated without a permanent leader since 2020. Its research has been halted and staff must now report to the White House. Lawmakers fear this gives the White House too much control over AI safety work. Representative Jamie Raskin called it an attempt to control a vital national security function. The center focuses on risks like bias and cybersecurity threats in AI systems.

Data centre anger hurts Trump midterm hopes over electricity bills

Opposition to data centres became a major issue in US midterm elections. Democrats used the issue in competitive races in Ohio, Michigan, South Carolina and other states. Residents worry that data centres will raise electricity bills due to costly infrastructure upgrades. The growing number of data centres has sparked debate over who should pay for the needed power and energy investments.

The hidden environmental cost of Artificial Intelligence

AI growth is causing serious environmental problems due to data centre expansion. These include land clearing, water and energy use, and higher demands on power grids. The author argues that tech infrastructure can lead to environmental destruction if communities are left out of decisions. Not all data centres are harmful, as impact depends on design, cooling and energy sources. The writer calls for clear rules to protect the environment and for companies to share accurate energy and water use data.

Google forces AI search on California schools sparking backlash

Google announced it will block K-12 students from using its regular search engine starting in January unless school districts opt into its AI-powered version. Los Angeles Unified and other districts are reviewing the change. Parent advocacy group Schools Beyond Screens called the move a hostile ultimatum. Common Sense Media said removing standard search limits student learning opportunities. Some districts may struggle to switch away from Google due to long contracts and costs.

AI design tools speed up work but reduce creativity study finds

A study of 34 fashion design students in South Korea found that using generative AI in Adobe Illustrator made textile design tasks faster but reduced creative autonomy. Researcher Shin Young Jang published the findings in the journal Fashion and Textiles. Students used Firefly-powered tools to generate designs for items like caps and bags. The study showed AI lowered cognitive workload but weakened personal style expression. Most students had used ChatGPT before but were new to Adobe Firefly.

AI investment boom may slow down as revenues lag spending

Capital Economics warned that the AI investment boom may become unsustainable as tech companies spend billions without matching revenue growth. The five largest US hyperscalers plan to spend $800 billion this year on AI infrastructure. AI investment growth is expected to slow in 2027 and fall about 10 percent in 2028. Annual AI revenues would need to jump from $150 billion to $200 billion now to $1.7 trillion to $1.9 trillion by 2030. The slowdown could hurt US and Asian economies tied to AI demand.

AI nearly led US troops to attack wrong target report reveals

A US aircraft nearly attacked a Chinese ship in the Middle East after an AI chatbot wrongly concluded it was carrying nuclear weapons components. An analyst used AI to reach the flawed conclusion and again to write the intelligence report. Officials canceled the operation only after closer review. The incident highlights a broader problem of AI systems producing no records that allow others to verify their reasoning. The author called for independent oversight and verification tools similar to aviation flight recorders.

New report compares tools for detecting unauthorized AI use at work

A new comparison examines how five security products detect Shadow AI, which is any AI tool employees use without IT approval. The report evaluates Sophos, CrowdStrike, Microsoft, Netskope and Zscaler across endpoint, network, browser and identity layers. Gartner now treats the endpoint as a key control point for Shadow AI. IDC projected 1.3 billion AI agents by 2028. The comparison covers 13 dimensions including local agent discovery and network enforcement based on vendor documentation from November 2026.

AI cyber resilience lags behind organizational confidence

A new study found that organizations are overconfident in their ability to recover from AI-related cyber incidents. Only four out of 25 organizations have practiced an AI or agent incident scenario. The research calls for more exercise of existing playbooks and human oversight. It highlights the need to validate tools across the full agent workflow and rehearse recovery to ensure safe business resumption.

Game developers told to adapt to AI clone problem

An opinion column argues that the video game industry is being harmed by the rise of AI. It points to rising costs for consoles and everyday goods. Independent developer NasNakarus saw over a dozen AI-assisted clones of their viral game appear on Steam. The author says game developers must adapt to this new AI-plagued future.

Sources

• AI safety center faces White House oversight questions
• Staff at the US center for AI safety are reporting to the White House, raising independence concerns
• Staff at the US center for AI safety are reporting to the White House, raising independence concerns
• AI, electricity bills and voter anger. How data centres tanked Trump's midterm hopes
• The hidden environmental cost of Artificial Intelligence
• Google's new AI 'ultimatum' sparks concern over tech giant's power in California schools
• Generative Artificial Intelligence Tools Accelerate Design Tasks but Diminish Creative Autonomy
• How sustainable is the AI investment boom after Thursday’s selloff?
• AI nearly had US troops attack a Chinese ship. Nobody could check its work.
• Shadow AI Detection: 4 Control Layers Compared [2026]
• AI cyber resilience and the trust gap
• COLUMN: Game developers told to 'adapt' to new AI-plagued future