Another Friday, another selection of AI news and resources to help you become more AI native. This week:
Open AI’s secret model
Anthropic explores building its own AI chips
OpenAI Shares a Practical GPT-5 Playbook
AI News
Open AI’s secret model. OpenAI is finalizing a new cybersecurity-focused product set for exclusive release to a select group of partners — mirroring Anthropic's restricted rollout of its "Mythos" model over autonomous hacking concerns. Experts warn models can now independently find software flaws and write exploits, threatening critical infrastructure. While restricted releases offer some protection, open-source equivalents are expected to surface within months regardless.
Anthropic explores building its own AI chips. Anthropic is in early-stage exploration of building its own AI chips amid a growing hardware shortage — though no formal plans or team are in place yet. Rivals Meta and OpenAI already have custom chip projects underway. Currently, Claude runs on Amazon Trainium, Google TPUs, and Nvidia GPUs, with a major Google TPU expansion expected online in 2027.
Meta launches Muse Spark. The first model from its newly formed Meta Superintelligence Labs, led by former Scale AI CEO Alexandr Wang. Born from a $14.3B overhaul after Llama fell behind rivals, it excels at visual STEM, coding, and hardware troubleshooting. An upcoming "Contemplating" mode uses parallel agents for complex reasoning, though mandatory Facebook/Instagram login is already raising privacy alarms.
OpenAI Shares a Practical GPT-5 Playbook for Startups
The guide breaks down four core steps: migrate, optimize, steer, and troubleshoot.
OpenAI's practical guide to building with GPT-5 breaks the process into four core steps:
Migrating to the Responses API for smarter state management and future-proofing
Optimizing prompts through evals and metaprompting
Steering behavior with new reasoning effort and verbosity controls
Troubleshooting common pitfalls like overthinking or tool overuse.
GPT-5 is highly steerable, but unlocking its full potential demands prompt refinement and the right infrastructure from the start.
That's it for this week. See you next Friday.
- Cam
PS: Want an AI audit and 12-month roadmap to accelerate your business without increasing risk? Click here.