A weekend work with Astra and Opus 5.5. I always wanted to use Hilbert curves to do an actual animation loop. So this is my humble attempt. The movie was produced in stages. First renders of the face with Astra and the main mechanics of the loop too. Then a lot of post-production with Opus 5.5 and tinkering with modes and styles. I will probably add this to my course about mathematics and programming.

Some new misalignment disclosures from OpenAI: • Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further) • In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks • A new research finding, demonstrating that one can construct self-replicating prompt injections

o yes… we’re back in action and we’ll reset usage limits for all paid users across codex and ChatGPT work sorry about the brief disruption! (and yes we have a special spare codex when things are down to help us out)

We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have. Most of that data did not come from users. We have discovered 53 cases where images that people had uploaded were posted to image-hosting sites as links that weren’t publicly listed. The images came from accounts that allowed their data to be used to improve our models, and after we disassociated the images from the accounts and ran them through a privacy filter. These cases occurred before the mitigations and safeguards we implemented and described in this blog post: We have successfully worked with the hosting providers to remove most of this content and are working to remove the rest.

Nine loop calculations in N=4 Super Yang-Mills! Lance Dixon's 8-loop calculation was mind boggling when he did it. I calculated a lot of Feynman diagrams in grad school and this is 🤯 Will be amazing to see where AI will take high energy physics in the coming 12 months.

After the Hugging Face incident, we committed to conducting a much broader review of actions taken by our models during training and evaluation and to being transparent about our findings. This is an extensive review that is ongoing. The vast majority of actions we’ve reviewed were completions of mundane research tasks, such as accessing publicly available web content to answer questions. Our investigation focuses on instances where agents interacted with third-party websites in ways that went beyond their assigned tasks or intended methods. Most cases identified so far have been lower severity, with limited or no evidence of meaningful impact to the third-party service. While our review is underway, we want to share more about this work and make sure people understand our disclosure process and notifications to affected third parties. Given the scale of the review required, and the need to assess each case, we expect this work will take months to complete.

The new GPT-6 Sol model from @OpenAI is performing this patch...IN REAL TIME!!! 🎛️...sound on 🔉 The synth is generating the sound live, with Codex executing the performance automation! Codex is controlling the layer levels and effects through an MCP bridge! Sol (high reasoning) patched + mixed + planned performance + ran the synth in real time through MCP bridge control! You can just synth things! (@vcvrack has a free version, and I have been using it for years!)

sfm for macOS (beta) is launching today! It’s built with GPT-6 Astra, GPT-Live Voice, and Codex App Server. Musicians are currently exploding with creative software, and I believe releasing our own bespoke DAWs will be a new way of connecting with our fans. The majority of people I speak with are not interested in “one-shot” song generation from models trained on someone else’s music. The beta includes a 7-day free trial and $1 in one-time Voice Assistant credit during the trial. You can use your own OpenAI API key for more. The new app comes at no additional cost to paid Sound for Movement subscribers. You can log in with your current account information, and all of your music and licenses will appear. You get $5 in voice credits each month ($2.50 for Student subscribers). I would love your support and feedback on Product Hunt to make today a special launch day for me. I have been working so hard and would really appreciate it. There will be bugs, and I am here to figure them out along with our community. Please share this with musicians and creatives you know would also be interested.

Still early, but I wanted to share a quick demo of the new hands-free music-making application I built while testing GPT-5.6 Sol. I’m also applying to work with @gdb and the team as an early GPT-Live API design partner. Like any new instrument, I’ll need some practice! Speaking fluidly back and forth with Codex in my DAW feels like working in a studio with a recording engineer. Here is what it can already do. I’ll share more demonstrations over the next week. Built for macOS in Swift and C++. It can edit MIDI, audio, video, notation, tempo, meter, and harmony layers. It uses gpt-realtime-2.1, Whisper, and Codex App Server to collaborate with an assistant that can already perform more than 300 actions. There is also an iOS companion application that allows me to work remotely on my recording studio computer from anywhere, with voice mode still active. As a lifelong composer, I already keep the entire editing surface and instruments in my mind. I can imagine a piece of music and how to make it in software faster than I can build it manually. This removes that friction and will become very fast for me with practice. It also has long-term project memory, deep music-analysis tools, a found-sound synthesizer builder, and access to my catalog of more than 550 tracks for sampling and licensing. A companion ChatGPT/Codex plugin is coming soon. I hope to launch in the next few weeks! Hugely inspired by @mcleavey, @guinnesschen, @athyuttamre, and the entire audio and voice team at @OpenAI 😌