OpenAI releases Codex ‘Chronicle’ feature for enhancing context, here’s how it works

Key Takeaway: OpenAI’s new Codex “Chronicle” feature helps the app remember what you were doing by using recent screen context, so you can give vaguer prompts like “this” or “that” and still get relevant help—though it comes with real privacy, security, and rate-limit tradeoffs.

More Insights:

• Chronicle expands Codex memory by learning from recent screen activity, not just chat history.

• It is designed to reduce how much context users have to manually repeat in prompts.

• The feature is available as an opt-in research preview for ChatGPT Pro users on macOS.

• It requires Screen Recording and Accessibility permissions to work.

• OpenAI warns that Chronicle can burn through rate limits quickly and increases prompt-injection and privacy risks because screen-derived memories are stored locally on device.

Why it matters: This matters because AI tools are moving beyond answering prompts and starting to understand ongoing work like a true assistant—but the more helpful they become, the more they blur the line between convenience and surveillance. Chronicle is a glimpse of the future of context-aware AI, and it raises the bigger question: how much of your digital life are you willing to let an AI watch in exchange for saving time?

Meet Otis — The World's First Cinematic AI Experience

Most AI puts you in front of a screen. Otis puts you on a porch.

He's a wise elder in a rocking chair watching a sunset — back turned, voice warm, ready to talk through whatever you're carrying. Career decisions. Hard conversations. Life advice from someone who has already walked your road.

No bullet points. No cold responses. Just presence.

The world has never seen an AI experience like this. Now it can.

Now live on Kickstarter — lifetime access available for founding members only.

Peter Thiel Is Building a Parallel Justice System – Powered by AI

Key Takeaway: This article argues that Peter Thiel’s backing of Objection.ai is not about truth or accountability, but about building a cheaper, privatized system to intimidate journalists and weaken press freedom outside democratic legal institutions.

More Insights:

• Objection ai lets people challenge media reports through private investigations and AI-generated verdicts instead of real courts.

• The article says this extends Thiel’s long-running hostility toward the press, dating back to the Gawker case.

• Its process appears designed to pressure reporters with ex-intelligence investigators, arbitration demands, and public “trust score” attacks.

• The author argues the company uses the language of facts and accountability while actually enabling quasi-legal harassment.

• In today’s political and media climate, the article warns this kind of private system could have a chilling effect far beyond the few cases it directly targets.

Why it matters: What makes this important is not just one startup or one billionaire’s grudge, but the larger idea behind it: when powerful people stop trying to win in public institutions and start building private systems of “truth” and “justice,” accountability itself becomes something the rich can buy and weaponize.

AI companies are buying the Slack data of failed startups

Key Takeaway: Defunct startups are turning their internal Slack threads, emails, Jira tickets, and other “operational exhaust” into valuable AI training data, creating a new market that helps failed companies recover cash while raising serious privacy concerns.

More Insights:

• Shutdown firms can now sell years of internal digital work for tens or even hundreds of thousands of dollars.

• AI labs want messy, real workplace data because public internet data is no longer enough for training useful AI agents.

• Startups like SimpleClosure and Sunset are building businesses around packaging and selling these archives.

• The most valuable datasets are deeply connected across tools, like Jira tickets linked to code commits.

• Even with anonymization, experts warn that employee communications may still expose sensitive identities or be reproduced by AI systems.

Why it matters: This is bigger than a clever liquidation strategy. It signals that the future of AI may be built not just on public knowledge, but on the private residue of human work. That should make readers pause: when everyday office chatter becomes raw material for machine intelligence, the line between company asset and personal digital history starts to disappear.

Other stuff

• Anthropic's Claude Mythos Launch Is Built on Misinformation

• I Stumbled Across My Boyfriend's ChatGPT and It Ended Our Relationship

• Sam Altman’s project World looks to scale its human verification empire. First stop: Tinder.

• Ad buyers say ad rates for ChatGPT are falling from $60 CPM to as low as $25 and the minimum spend to advertise is down to $50K from $250K at launch

• China has nearly erased US lead in AI&the flow of experts moving to US is slowng

• DARPA builds AI to investigate China's claim it can break military encryption

• Guy builds AI driven hardware hacker arm from duct tape, old cam and CNC machine

Hope you enjoyed today's newsletter

⚡️ Join over 300,000 people using the Superpower ChatGPT extension on Chrome and Firefox.