Sponsored by

ChatGPT probably isn’t conscious. But what if we’re wrong?

Key Takeaway: The article explores whether AI systems like ChatGPT or Claude could ever become conscious, arguing that the debate depends less on current chatbot behavior and more on unresolved questions about what consciousness actually is.

More Insights:

• Some major AI thinkers believe today’s chatbots may already have some form of subjective experience.

• The strongest argument for AI consciousness comes from “computational functionalism,” the idea that consciousness could emerge from information processing.

• Skeptics argue that bodies, biology, emotions, and sensory experience may be essential for true consciousness.

• The article warns that treating non-conscious machines as sentient could make people easier to manipulate or afraid to shut AI systems down.

• It concludes that AI consciousness deserves study, but existing animal suffering is a far more urgent moral issue.

Why it matters: The AI consciousness debate forces us to confront a deeper question: are we building tools, or are we accidentally creating beings we could harm? The danger is not just that we might ignore future digital suffering — it’s that we still struggle to care about conscious suffering already happening around us.

Free Weekly AI Sessions for Experienced Software Engineers.

Every Wednesday at 5 PM CT, Gauntlet AI professors teach a live, hands-on AI engineering session — completely free. If you're nontechnical, this isn't for you. New topic every week, built for engineers who want to build, not just watch. See upcoming sessions.

Browse Free Sessions

AI law firm wins UK court case for first time

Key Takeaway: An AI-powered UK law firm helped a freelancer win a £7,000 unpaid-fees case, showing how legal tech could make small claims cheaper and easier to pursue.

More Insights:

• Garfield AI prepared the legal documents, including witness statements, while a human barrister handled the courtroom advocacy.

• The freelancer paid about £400 to recover £7,000 from a hospitality business.

• Garfield AI was approved by the UK Solicitors Regulation Authority last year.

• The firm has processed over 600 claims and recovered around £500,000 for clients.

• The case highlights AI’s promise in routine legal disputes, despite recent AI-related legal mistakes by major firms.

Why it matters: This could be the beginning of a major shift in access to justice: AI may not replace courts or lawyers, but it can make legal action possible for people who would otherwise walk away because the process is too expensive or intimidating.

Meta pauses an AI training program that tracks employees' keystrokes after an internal leak

Key Takeaway: Meta paused an internal AI training program that records employees’ keystrokes and mouse movements after sensitive staff data was accidentally made accessible across the company.

More Insights:

• The leaked data reportedly included private conversations, performance data, and transcriptions.

• Meta classified the incident as a SEV 2, meaning it was serious but not the most severe level.

• The AI program, called Model Capability Initiative, was mandatory for most employees.

• Employees had already raised concerns about being monitored through keystrokes and mouse movements.

• Meta says it has no evidence that employees improperly accessed the exposed data.

Why it matters: This shows the growing tension between AI development and workplace privacy; even companies building the future of AI can lose trust quickly when employees feel they are being turned into training data without strong safeguards.

Other stuff

• The Korean telecom giant at the center of Anthropic's Mythos controversy

• 28 Tips to Take Your ChatGPT Prompts to the Next Level

• How Anthropic may have talked itself into an AI export ban

• Record & Replay – Codex

• Codex-maxxing for long-running work

• Designing delightful front ends with GPT-5.4

• Tech Workers Who Don’t Embrace AI Face Triple the Layoff Risk, Gallup Finds

• MosaicLeaks: Can your research agent keep a secret?

• Applications are open for OpenAI DevDay 2026

• My coworkers read my personal ChatGPT chats via Meta Analytics [video]

Hope you enjoyed today's newsletter

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