US uses Artificial Intelligence to spot child abuse images generated by Artificial Intelligence, plus climate tech list preview

Homeland security investigators are piloting software to tell images generated by Artificial Intelligence from those with real victims, while MIT Technology Review tees up its 2025 Climate Tech Companies to Watch. The newsletter also rounds up major developments across safety, energy, and robotics.

Generative tools have driven a surge in child sexual abuse imagery, and the leading US investigator of child exploitation is now testing whether Artificial Intelligence can help. A new government filing reveals the Department of Homeland Security’s Cyber Crimes Center, which pursues cross-border exploitation cases, has awarded a contract to San Francisco-based Hive AI for software that assesses whether content was generated by Artificial Intelligence. The effort aims to distinguish synthetic images from material depicting real victims, potentially helping investigators target resources and evidence gathering more effectively.

MIT Technology Review will publish its third annual Climate Tech Companies to Watch on October 6, highlighting firms with the potential to curb emissions and help communities adapt. The urgency is underscored by another year of record temperatures and climate-fueled disasters costing billions worldwide. Building on two prior editions, this year’s list, previewed by Casey Crownhart, follows the 2024 roundup and seeks to spotlight practical innovations that can move the needle on warming and resilience.

Today’s link roundup covers safety, spending, and next-gen models in Artificial Intelligence. ChatGPT’s new parental controls can alert parents and law enforcement when users under 18 discuss self harm or suicide, debuting amid pressure on chatbot makers to bolster safeguards. Companies’ Artificial Intelligence budgets continue to balloon with uncertain returns. Energy stories include concern among oil executives about attacks on offshore wind and how states could still advance clean power. Urban fire prevention highlights the importance of making homes fire-resistant, with separate work on using Artificial Intelligence to detect wildfires. Top firms are betting on world models trained on video and robotics data, and China is racing to electrify freight trucks. Other pieces examine the loss of nuance in Artificial Intelligence translation, the ship-cleaning robots of Neptune Robotics, and a flurry of attention on an Artificial Intelligence actress. A quoted venture capitalist notes intense startup pressure in an era of Artificial Intelligence hype.

The “one more thing” centers on DeepSeek’s rise as a homegrown challenger to OpenAI and its unexpected role in reviving fortune-telling among young people in China. Users are sharing Artificial Intelligence-generated readings, crafting fortune-telling prompts, and revisiting spiritual texts through chatbots. The trend reflects broader social anxiety and the limits on public religious practice, with screens offering a private outlet. The edition closes with lighter diversions, including a site to revisit the news from your birthdate, a rock track by Silica Gel, an interactive tour of Bosch’s Garden of Earthly Delights, and a global map of movie settings.

63

Impact Score

Why DeepSeek v4 matters

DeepSeek’s new open-source flagship pairs stronger performance with a much longer context window and early support for domestic Chinese chips. The release signals progress in open models, memory efficiency, and China’s push to reduce reliance on Nvidia.

OpenAI launches workspace agents in ChatGPT

OpenAI has introduced workspace agents in ChatGPT, giving teams shared Codex-powered agents that can handle multi-step work across business tools and Slack. The feature is aimed at recurring organizational workflows with admin controls, approvals, and enterprise monitoring.

Generative Artificial Intelligence in B2B sales and content creation

Generative Artificial Intelligence is presented as a way to reduce inefficiencies in customer-facing sales work and the production of sales materials. The research combines literature review, survey data, and a pilot experiment to identify where gains are most practical in B2B sales environments.

ChatGPT Images adds thinking capability

OpenAI has upgraded ChatGPT Images with a new thinking mode that can search the internet, generate multiple images, and verify outputs before finalizing results. The update also improves text rendering, dense compositions, multilingual support, and style flexibility.

OpenAI launches workspace agents in ChatGPT

OpenAI has introduced workspace agents in ChatGPT, giving teams shared Codex-powered agents that can handle multi-step work across business tools and Slack. The feature is aimed at recurring organizational workflows with admin controls, approvals, and enterprise monitoring.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.