NYT Fires Writer Over AI Copying. AI Cyber Risk Accelerates. Microsoft Labels Copilot Entertainment. Gemini Guides a City Day.

Hey, and welcome to the AI Briefing Daily, your podcasts for the most important AI news from the past 24 hours.

Today is April 6th, 2026, and here are the latest AI updates.

NYT fires writer over AI copying.

AI Cyber Risk accelerates.

Microsoft labels co-pilot entertainment.

Gemini guides a city day.

This episode is sponsored by RocketRoutine, run your company with an AI operating system built for founders who want to move fast without losing control.

The New York Times cut ties with freelance journalist Alex Preston after an AI tool he used copied passages from a Guardian review into his New York Times book review of watching over her by Jean-Baptiste Andrea.

A reader spotted the overlap and the paper ended the collaboration.

Preston said he was ashamed and called it a mistake.

The incident suggests the AI pulled directly from an earlier Guardian piece and that Preston failed to catch it.

A similar error at R's Technica saw fabricated quotes from chat GPT after an editor relied on its internet crawling and did not verify sources.

The cases underscore the need to check AI outputs.

AI models offensive cyber capability is rising quickly doubling roughly every 9.8 months since 2019 and every 5.7 months since 2024 liptus research reports.

10 security experts evaluated 291 tasks opus 4.6 and GPT 5.3 codex with a token budget of 2 million achieved 50% success on tasks that take human experts 3 hours.

The larger budgets performance jumps GPT 5.3 codex goes from 3.1 to 10.5 hours at 10 million tokens suggesting actual attack risk may be underestimated.

Open source models lag closed models by 5.7 months data are on GitHub and hugging face.

Your company has a strategy but does it actually reach your team every single day.

Most founders spend more time managing the system than running the business rocket routine OS connects goals to daily execution using AI operators that work while you focus on what matters less firefighting more output join the waitlist at rocket dash routine dot com.

Microsoft's co-pilot is described in its terms of use as for entertainment purposes only the terms apparently last updated on the 24th of October 2025 warn that it can make mistakes and it may not work as intended until users not to rely on co-pilot for important advice and to use co-pilot at your own risk.

A Microsoft spokesperson told PC mag the phrasing is legacy language and will be changed in the next update.

Tom's hardware notes similar disclaimers from open AI and XAI who also caution that their outputs should not be treated as unquestionable truth.

Gemini now lives inside Google maps as ask maps and can plan a day out I asked for a public transit route with lunch a walk and a laptop friendly coffee stop across two neighborhoods.

Home by 430 it suggested tacos chookies a detour to a bookstore that mislocated the walk but corrected to cobo and volunteer park conservatory where admission was $6 and a warm greenhouse was perfect in the rain.

The day ended at day made cafe and I reached home at 426 Gemini polls reviews weather and maps data shows its sources but can hallucinate people supply the recommendations Gemini stitches them together.

That was the AI news of the day with this you are perfectly informed for today.

Would you like to create an automated podcast on your own topic?

Get in touch with us and we'll make it happen.

Thank you for tuning in. See you tomorrow for the next AI briefing.