Hallucination Station: Issue 10

🌟 Editor's Note

Double digits baby! Congrats on joining the community! If you like what you see then please share with your product friends so we can iterate and improve and bring you the best newsletter possible!

🚀 This Weeks Top Stories in AI

🚀 OpenAI’s Dev Day 2025: A New Chapter for ChatGPT

• OpenAI launched in-ChatGPT Apps (e.g. Canva, Spotify) plus AgentKit tools for building agent workflows, essentially turning ChatGPT into a conversational OS.

• They opened API access to stronger models (GPT-5 Pro, Sora 2, realtime voice models) and made GPT-5-Codex generally available with enterprise controls.

• OpenAI struck a compute deal with AMD (6 GW commitment, up to 10 % stake).

👉 Read More About It

🤝 Deloitte + Anthropic: Scaling Trusted AI Together

• Deloitte will roll out Anthropic’s Claude across its 470,000 employees and build industry-specific solutions in regulated sectors using its Trustworthy AI framework.

• The partnership includes launching a Claude Center of Excellence and training/certification for 15,000 Deloitte professionals to ensure best practices in AI deployment.

• Early applications include AI Advantage for CFOs, automating finance workflows and insights, combining Claude, AWS, and Deloitte’s domain know-how.

👉 Read More About It

🤖 OpenAI Agent Builder: Visual AI Workflows Made Easy

• Drag-and-Drop Workflow Design: Agent Builder offers a visual canvas for creating and versioning multi-agent workflows, allowing developers to design logic with drag-and-drop nodes and configure custom guardrails.

• Seamless Integration with Tools: The Connector Registry enables admins to manage data and tool connections across OpenAI products, supporting integrations with services like Dropbox, Google Drive, and Microsoft Teams.

• Enhanced Safety and Performance: Developers can implement Guardrails to protect agents from unintended behavior and utilize Evals capabilities, including datasets, trace grading, and automated prompt optimization, to measure and improve agent performance.

👉 Read full breakdown

🧠 MIT Study: AI Use Linked to Reduced Brain Activity

• A recent MIT study found that users relying on AI tools like ChatGPT showed decreased brain activity during cognitive tasks, indicating reduced mental engagement.

• Participants using AI assistance produced essays with less neural and linguistic variation, suggesting a decline in critical thinking and creativity.

• The study also observed a drop in "ownership" of work, with AI users struggling to recall or quote their own writing, highlighting potential impacts on learning and memory retention.

👉 Read Full Breakdown

🦄Founder Corner

• Bezos bets AI is a Bubble

• Questions every AI Founder should be asking

• 20-Year-Old Indian founder secures $3M for Supermemory

🔥Meme-while in AI

(Coming next week - send me your favorite AI memes - [email protected] - also please please please share your feedback!)

🏆Tool of the Week (not sponsored)

For the week that’s in it… why not line Altman’s coffers some more:

Did You Know? I have been banned from Twitter twice, despite having never posted, commented, or engaged with anything or anyone on the platform….go figure!

Till next time,