# Employed AI > Independent reporting on real AI use cases, the tools people used, and what happened. ## Explore the publication - [Homepage](https://employed-ai.com/): The latest reporting and practical AI use cases. - [Article archive](https://employed-ai.com/archive/): Every published article in chronological order. - [About and editorial approach](https://employed-ai.com/about/): Scope, bylines, and how articles credit their sources. - [RSS feed](https://employed-ai.com/rss.xml): Subscribe to new articles. ## Topics - [AI Agents](https://employed-ai.com/ai-agents/): How AI agents use tools, carry out tasks, and report their work. Follow the steps between a request, an action, and an outcome you can verify. - [AI Use Cases](https://employed-ai.com/ai-use-cases/): See how people used AI to review code, look up clinical evidence, and run laboratory experiments. Each example explains what happened and what people checked. - [AI Tools and Software](https://employed-ai.com/ai-tools/): AI product releases, demos, and workflows. Find out what you can try, what access or setup it requires, and what evidence you have for choosing a tool. - [AI Research](https://employed-ai.com/ai-research/): Read AI papers, benchmarks, and demonstrations with an eye on what was tested, how results were scored, and what someone else could verify. - [AI Security](https://employed-ai.com/ai-security/): How AI tools handle access, credentials, and private data. Follow incidents and tests to see what crossed a security boundary and what controls could catch it. ## Recent articles - [Antigravity’s one-hour chemistry workflow: what Google’s demonstration shows](https://employed-ai.com/articles/antigravity-one-hour-chemistry-workflow/): Google reports using Antigravity to curate compound data, compare molecular structures and propose new designs in an hour. Here is what those outputs establish and what scientists still need to check. - [How Souffleur turns AI voice conversations into glanceable topic cards](https://employed-ai.com/articles/how-souffleur-turns-ai-voice-conversations-into-glanceable-topic-cards/): Souffleur is a research prototype that displays voice chats as topics, decisions and short gists. Its design shows what a compact overview can leave out. - [How Maike is designed to connect AI writing dialogue with student revision](https://employed-ai.com/articles/maike-ai-writing-dialogue-student-revision/): Maike's proposed writing-tutor architecture links essay diagnosis, targeted questions and revision planning to edits students write themselves. - [Sudan responders report uses for generative AI, and limits that need local judgment](https://employed-ai.com/articles/sudan-responders-report-generative-ai-uses-limits-local-judgment/): An exploratory preprint describes how Emergency Response Room volunteers used chatbots, checked suggestions with people and weighed sensitive-data risks. - [A preprint proposes training-data backdoors for multi-turn AI dialogue](https://employed-ai.com/articles/training-data-backdoors-multi-turn-ai-dialogue/): A proposed attack framework uses poisoned fine-tuning conversations to target later dialogue. It assumes eligible colluding contributors and acceptance of their records. - [AI-generated maths proofs: what does ‘verified’ mean?](https://employed-ai.com/articles/ai-generated-maths-proofs-what-does-verified-mean/): OpenAI’s October 2026 disclosures and corrections show why checking an AI-generated argument, establishing novelty and understanding a result require different work. - [AI-invented citations and research misconduct: what the policies establish](https://employed-ai.com/articles/ai-invented-citations-research-misconduct/): arXiv assigns authors responsibility for incorrect references, while ICMJE recommends source checks. Whether an invented citation constitutes formal misconduct remains unresolved here. - [How to read an AI text-watermark robustness test](https://employed-ai.com/articles/how-to-read-an-ai-text-watermark-robustness-test/): A worked example shows how attacker access, test prompts, quality checks and detector calibration shape what a watermark robustness study can establish.