Wiredframe Radar logo

Category intelligence

Social Media Briefing — June 23, 2026

433 current items analyzed and ranked.

Executive synthesis

Social Media Summary

OpenAI's Daybreak cybersecurity launch dominated the day, led by Sam Altman, Greg Brockman, and the official OpenAI account.

  • OpenAI unveiled the full GPT-5.5-Cyber model, Codex Security, Patch the Planet, and a Cyber Partner Program with Trail of Bits and HackerOne, positioning frontier AI for defensive vuln-finding and patching at machine speed.
  • Sakana AI's David Ha (hardmaru) launched Fugu, an orchestration model coordinating multiple frontier models, framed as a sovereignty hedge against vendor lock-in. Ethan Mollick offered skeptical hands-on tests, calling Fugu Ultra slow and unable to match the Fable model.

Research and architecture debates ran parallel. Nathan Lambert introduced TMax, an open RL recipe for terminal agents, and separately hailed GLM-5.2 as a "DeepSeek moment" for open agentic models. LangChain's Harrison Chase distinguished model routing from model councils, a trending design topic.

Key Themes

OpenAI Cybersecurity Push · 4OpenAI Daybreak Cyber Launch · 5Sakana Fugu Launch and Orchestration Models · 11RL for Agents / TMax · 4AI Adoption and Public Sentiment · 19Open Models & GLM-5.2 · 2AI-Kills-SaaS Debate and Adobe Defense · 13Sakana Fugu Launch · 10Model Routing vs Model Council · 2AI Infrastructure and Compute Economics · 5

Primary evidence

Top Ranked Signals

Social Twitter Jun 22

We want to help all companies be secure, working with the USG and the security ecosystem. *The full...

By @sama

85 score
AI Analysis

Sam Altman announces the full GPT-5.5-Cyber with state-of-the-art CyberGym performance, plus Patch the Planet and Codex Security to fix not just find security problems.

We want to help all companies be secure, working with the USG and the security ecosystem. *The full version of GPT-5.5-Cyber is here; state of the art performance on CyberGym. *Patch The Planet and Codex Security will help solve security problems instead of just finding them. t.co/otyCFHJR4d
AI securityOpenAImodel releaseGPT-5.5-Cyber
Social Twitter Jun 22

Human intelligence is fundamentally a collective intelligence. We solve complex problems by particip...

By @hardmaru

80 score
AI Analysis

Hardmaru launches Sakana Fugu, an orchestration model that dynamically coordinates the world's best models to match restricted frontier models like Fable and Mythos, framing orchestration as the next frontier beyond bigger models and collective intelligence.

Human intelligence is fundamentally a collective intelligence. We solve complex problems by participating in a vast cultural network that builds upon ideas across generations. I believe the strongest AI systems will become a collective intelligence, too. Since we started Sakana AI, our core conviction has been that the most powerful AI systems will be collaborative ecosystems, not isolated monoliths. Evolution innovates under constraints, and the future belongs to systems that explicitly learn
Sakana Fugumodel orchestrationcollective intelligencemodel release
Social Twitter Jun 22

We’re expanding OpenAI Daybreak to help democratize patching vulnerable software at machine speed: ...

By @OpenAI

76 score
AI Analysis

OpenAI's headline post expanding Daybreak with the Codex Security plugin, the full GPT-5.5-Cyber model, a Cyber Partner Program, and Patch the Planet to scale defensive patching at machine speed.

We’re expanding OpenAI Daybreak to help democratize patching vulnerable software at machine speed:
  • Codex Security plugin: find, validate, and fix vulnerabilities right inside Codex
  • The full version of GPT-5.5-Cyber model: a great model for trusted defenders
  • Cyber Partner Program: powering products built on top of our best cyber capabilities for leading security companies to secure the world's software
  • Patch the Planet: working with maintainers to secure critical open source projects
OpenAI cybersecurityAI security toolingproduct launch
Social Twitter Jun 22

TMax: An open RL recipe for terminal agents I’m very excited to get to share a new RL paper today t...

By @natolambert

75 score
AI Analysis

natolambert introduces TMax, an open RL recipe and dataset for terminal agents, reflecting on how RL research in mid-2026 has shifted from the RLVR math era toward harder agentic tool-use tasks requiring greater empirical rigor.

TMax: An open RL recipe for terminal agents I’m very excited to get to share a new RL paper today that I got to have a small part in – a type of paper I suspect we’ll see much more of in the future. The key is that RL research is very different today, in mid-2026, than what most observers have in their context. The average conception of an RL paper is grounded in the RLVR revolution of early 2025, where many people could use vanilla RLVR libraries to hillclimb on math benchmarks. Crucially, thi
RLagentsopen-researchterminal-agents
Social Twitter Jun 22

GPT-5.5-Cyber is our most capable cyber model yet, designed for advanced, authorized defensive work:...

By @OpenAI

72 score
AI Analysis

OpenAI presents GPT-5.5-Cyber as its most capable cyber model for authorized defensive work including tracing vulnerable code, validating issues, and developing patches for human review.

GPT-5.5-Cyber is our most capable cyber model yet, designed for advanced, authorized defensive work: tracing vulnerable code, validating issues, developing patches, and preparing evidence for human review. t.co/KcDoGGD2tx
OpenAI cybersecuritymodel releaseAI security tooling
Social Twitter Jun 22

have been thinking a bunch about model routing and related things current thoughts here, would love...

By @hwchase17

72 score
AI Analysis

hwchase17 shares detailed thoughts distinguishing model routing (cost-driven, route to one best model) from model council (parallel multi-model aggregation for frontier performance), referencing OpenRouter Fusion and Sakana fugu, and debates feasibility of routing for everyday users.

have been thinking a bunch about model routing and related things current thoughts here, would love feedback: 1/ there is a difference between "model routing" and "model council" "model routing" = route to one best model. i think only real reason to do this is cost "model council" = pass to many models at the same time, then aggregate responses. this is done to push frontier performance 2/ "model council" seems to be having a moment recent. @OpenRouter did their Fusion thing, @SakanaAILabs
model-routingagentsinference-architecture
Social Twitter Jun 22

Anti-data center yard signs are popping up in my area. I am entertaining the idea of paying for a bi...

By @ID_AA_Carmack

72 score
AI Analysis

John Carmack notes anti-data-center yard signs appearing in his area and jokes about funding a pro-data-center billboard in Texas.

Anti-data center yard signs are popping up in my area. I am entertaining the idea of paying for a billboard with something like “Data centers are awesome, Texas should lead!”
data centerspublic sentimentAI infrastructure
Social Twitter Jun 22

We're accelerating patching, in addition to vuln finding, with new tools and models in OpenAI Daybre...

By @gdb

70 score
AI Analysis

Greg Brockman says OpenAI Daybreak models are discovering and generating patches for critical vulns in browsers, network infrastructure, and OSes like FreeBSD and Linux, and patching projects such as cURL, Go, Python, and cryptography libraries.

We're accelerating patching, in addition to vuln finding, with new tools and models in OpenAI Daybreak. Our models are now discovering and generating patches for critical vulns in major browsers, network infrastructure, and operating systems (such as FreeBSD and the Linux kernel), and patching projects like cURL, Go, Python, Sigstore, and pyca/cryptography. Working together with partners and the ecosystem to help secure the world's software:
AI securityOpenAIopen sourcevulnerability patching
Social Twitter Jun 22

Patch the Planet is our effort to help open source maintainers move from security findings to merged...

By @OpenAI

70 score
AI Analysis

OpenAI introduces Patch the Planet, working with Trail of Bits, HackerOne and maintainers to bring Codex Security and advanced models into open-source vulnerability remediation with human review.

Patch the Planet is our effort to help open source maintainers move from security findings to merged fixes. We’re working with Trail of Bits, HackerOne, Calif, researchers, and maintainers to bring Codex Security and advanced models into the remediation process, with human review at the center.
OpenAI cybersecurityopen sourceAI security tooling
Social Twitter Jun 22

Water usage has been a hot topic in the AI data center world, but the numbers may surprise you. Acc...

By @nvidia

70 score
AI Analysis

NVIDIA argues AI data centers use only about 0.2 percent of US daily water and that 45C liquid cooling can cut cooling water from roughly 2.6M gallons per MW per year to near zero, enabling heat reuse.

Water usage has been a hot topic in the AI data center world, but the numbers may surprise you. According to the Manhattan Institute, data centers use 0.2 percent of daily water usage in the U.S. and that number has dramatically decreased in the past few years due to a new method: liquid cooling. By moving to 45°C liquid cooling, AI factories in favorable climates can use dry coolers instead of conventional cooling-tower-based systems, cutting facility cooling water use from roughly 2.6M gall
AI data centerssustainabilityliquid coolingAI infrastructure
Social Twitter Jun 22

GLM-5.2 should be “DeepSeek moment” for agents. We enter a new world where the top end of agentic ca...

By @natolambert

70 score
AI Analysis

natolambert frames GLM-5.2 as a DeepSeek moment for agents, claiming top-tier agentic capability is now in open models and urging the open-source community to inform regulators on safe frontier open intelligence.

GLM-5.2 should be “DeepSeek moment” for agents. We enter a new world where the top end of agentic capabilities are available in open models. If you care about open, now is the time to inform regulators on how we should build a world with safe, frontier, open intelligence.
open-modelsagentsai-policyGLM
Social Twitter Jun 22

I thought “no way is this real”, but there actually is a project along these lines. I doubt Nvidia i...

By @ID_AA_Carmack

70 score
AI Analysis

John Carmack reacts to a pitch for distributed Nvidia-powered data centers, arguing it is economically nonsensical given large data centers' efficiencies of scale even for low-grade inference.

I thought “no way is this real”, but there actually is a project along these lines. I doubt Nvidia is any more involved than just selling them GPUs, and they aren’t going to be “silent”, but a company is actually pitching this. Large data centers have substantial efficiencies of scale, so this is economically nonsensical even for low grade inference (to say nothing of training) unless traditional construction is outlawed. t.co/v0tbB8oaFl
data center economicsAI infrastructureinferenceGPUs