AI Radar: Daily Tech Brief

In this episode of AI RADAR, we explore why **OpenAI is officially discontinuing Sora**, shutting down the expensive video platform. We also analyze Sam Altman's new **Our Principles** manifesto, aimed at balancing military demands with AI democratization. Plus, MIT and IBM Watson reveal **EnergAIzer** to rapidly predict and reduce AI's carbon footprint. Finally, brief updates on Google Earth Engine's compute tiers and the US Air Force's AI test assistant.

Sources:
• Sora Shutdown: https://help.openai.com/en/articles/what-to-know-about-the-sora-discontinuation
• Our Principles: https://openai.com/our-principles
• MIT EnergAIzer: https://news.mit.edu/2026/faster-way-estimate-ai-power-consumption-0427
• Google Earth Engine: https://developers.google.com/earth-engine/guides/noncommercial
• USAF AI Tool: https://www.aerospaceglobalnews.com/news/us-air-force-ai-tool-cuts-flight-test-planning-from-days-to-minutes/

On AI RADAR, we break down **OpenAI's GPT-5.5**, a new "AI coworker" with a 1-million-token context window. We also explore Google's staggering **$40 billion investment into Anthropic**, the largest single-company AI rival investment ever. Plus, **DeepSeek V4** debuts, disrupting the frontier market with a highly efficient open-weight model. Also: **Meta lays off 8,000 employees** to fund its generative AI push, and **Anthropic adds persistent memory** to Claude managed agents.

Sources:
• OpenAI Introducing GPT-5.5: https://openai.com/index/introducing-gpt-5-5/
• Google to Invest in Anthropic: https://www.bloomberg.com/news/articles/2026-04-24/google-to-invest-up-to-40-billion-in-ai-startup-anthropic
• DeepSeek-V4-Pro: https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro
• Meta to Cut 8000 Jobs: https://www.cbsnews.com/news/meta-to-cut-8000-jobs-as-it-charges-into-ai/
• Anthropic Adds Memory: https://sdtimes.com/ai/anthropic-adds-memory-to-claude-managed-agents/

In this episode of AI RADAR, we explore OpenAI's surprise launch of **GPT-5.5**, pushing us closer to an AI super app with autonomous workflows. We also cover the launch-day breach of Anthropic's **Mythos model**, which early testers are calling overhyped. Then, a watershed moment: Alphabet confirms **50% of Google's code is now AI-generated**. Finally, quick hits on SpaceX's rumored **$60B Cursor buyout option** and **Gitar's $9M raise** to manage AI code sprawl.

Sources:
• GPT-5.5: https://techcrunch.com/2026/04/23/openai-releases-gpt-5-5-bringing-company-one-step-closer-to-an-ai-super-app/
• Mythos Leak: https://siliconangle.com/2026/04/22/anthropic-investigates-unauthorized-access-restricted-claude-mythos-ai-model/
• Google AI Code: https://www.computerweekly.com/news/Half-of-Googles-software-development-now-AI-generated
• SpaceX & Cursor: https://www.therundown.ai/p/anthropic-mythos-leaked
• Gitar Raise: https://www.computerweekly.com/blog/CW-Developer-Network/Gitar-tunes-up-to-tighten-strings-on-AI-generated-code

In this episode of AI RADAR, we dive into OpenAI's massive release of **GPT-5.5** and its new Pro tier, setting a new benchmark for autonomous coding and multi-step agentic workflows. We also discuss Anthropic's **Claude Mythos**, a model so advanced at autonomous hacking that it triggered global cybersecurity alarms and forced a pause on its public release. Finally, we explore how **NVIDIA** and **Google Cloud** are teaming up to build "AI factories" powered by Vera Rubin GPUs to push physical AI into real-world industrial production. Plus, quick updates on GPT-5.5 shattering legal AI benchmarks and reclaiming the enterprise "workhorse" crown!

Sources:
• OpenAI GPT-5.5: https://openai.com/index/gpt-5-5-system-card/
• Claude Mythos: https://www.theguardian.com/technology/2026/apr/22/what-is-mythos-ai-and-why-could-it-be-a-threat-to-global-cybersecurity
• NVIDIA & Google Cloud: https://www.hpcwire.com/off-the-wire/nvidia-and-google-cloud-collaborate-to-advance-agentic-and-physical-ai/
• Harvey Legal AI: https://www.harvey.ai/blog/gpt-5-5-research-preview-results
• GPT-5.5 Workhorse: https://every.to/vibe-check/gpt-5-5-is-openai-s-workhorse-model

Welcome to AI RADAR. Today, we unpack a monumental day in frontier AI. First, **SpaceX** stuns the tech world with an option to acquire AI coding startup **Cursor** for **$60 billion**, giving it access to the Colossus supercomputer. Meanwhile, **OpenAI** shifts its enterprise strategy by pouring $1.5B into **DeployCo**, a $10B joint venture with private equity giants to embed AI directly into their portfolios.

We also cover **Anthropic** taking the Pentagon to court over strict AI safety guardrails and autonomous weapons. Plus, updates on OpenAI's new drug discovery model, **GPT-Rosalind**, and Alibaba's surprisingly powerful open-weight **Qwen3.6-27B**.

Sources:
• Quartz: https://qz.com/spacex-secured-an-option-to-buy-ai-coding-startup-cursor-for-60-billion
• TNW: https://thenextweb.com/news/openai-to-commit-up-to-1-5b-of-its-own-capital-to-deployco-a-joint-venture-with-pe-firms-valued-at-10b
• WaPo: https://www.washingtonpost.com/business/2026/04/22/anthropic-seeks-to-debunk-pentagons-claims-about-its-control-over-ai-technology-in-military-systems/
• VentureBeat: https://venturebeat.com/ai/openai-debuts-gpt-rosalind-a-new-limited-access-model-for-life-sciences-and-broader-codex-plugin-on-github/
• HuggingFace: https://huggingface.co/Qwen/Qwen3.6-27B

In this episode of AI RADAR, **Amazon** commits a staggering **$25 billion** to **Anthropic**, cementing their AWS alliance to rival Microsoft and OpenAI. We also explore the severe cyber threats posed by Anthropic's unreleased **"Mythos"** model. Plus, **Broadcom** strikes landmark custom AI chip deals with **Meta and Google**, and **OpenAI** faces a criminal probe in Florida.

Sources:
• Amazon/Anthropic: https://siliconangle.com/2026/04/21/amazon-to-invest-up-to-25b-in-anthropic-as-part-of-expanded-cloud-partnership/
• Mythos: https://siliconangle.com/2026/04/21/mythos-remains-a-mystery-as-security-world-faces-rising-threats/
• Broadcom Deals: https://blockonomi.com/broadcom-avgo-stock-rallies-21-on-massive-meta-and-google-ai-chip-partnerships/
• FL Probe: https://www.reuters.com/technology/artificial-intelligence/florida-launches-criminal-probe-into-openai-chatgpt-over-deadly-shooting-2026-04-21/
• OpenAI Policy: https://www.kalw.org/bay-area-news/2026-04-21/openai-drops-policy-proposal-for-the-intelligence-age

Today on AI RADAR: **Amazon's massive $100B infrastructure deal** with Anthropic, fueled by a new $5B investment to rival the Microsoft-OpenAI juggernaut. Next, **OpenAI's GPT-5.4-Cyber** launches exclusively for verified security defenders, shifting the AI safety paradigm. We also dissect OpenAI's **"Chronicle" for Codex**, a Mac feature capturing continuous screen context that sparks major enterprise privacy concerns. Plus: Google's strategic pivot to **Agentic AI** infrastructure and Anthropic's new global data center leases.

Sources:
• Amazon Invests in Anthropic: https://www.japantimes.co.jp/news/2026/04/21/business/corporate/amazon-to-invest-an-additional-5-billion-in-anthropic/
• GPT-5.4-Cyber Launch: https://www.marktechpost.com/2026/04/20/openai-scales-trusted-access-for-cyber-defense-with-gpt-5-4-cyber-a-fine-tuned-model-built-for-verified-security-defenders/
• Codex Chronicle Preview: https://9to5mac.com/2026/04/20/openai-releases-codex-chronicle-feature-for-enhancing-context/
• Google Cloud Next Agentic AI: https://siliconangle.com/2026/04/20/google-cloud-next-2026-preview-the-real-story-isnt-ai-its-the-control-plane/
• Anthropic Data Center Leases: https://www.datacenterdynamics.com/en/news/anthropic-seeks-data-center-leasing-deals-in-europe-and-australia/

On this AI RADAR, we explore **Meta's massive pivot**, cutting 8,000 jobs to fund $135B in AI infrastructure. We unpack **Stanford's 2026 AI Index Report** on the shrinking US-China AI gap and the boom of autonomous agents. Plus, the geopolitical clash at **NeurIPS**, where a ban on Chinese scientists was reversed. Finally, updates on **xAI's Grok Speech APIs** and **Microsoft Copilot**.

Sources:
• Meta: https://www.mlq.ai/meta-reportedly-to-cut-8000-jobs-in-upcoming-layoffs-amid-ai-cost-pressures/
• Stanford: https://thedecoder.com/stanfords-ai-index-2026-shows-rapid-progress-growing-safety-concerns-and-declining-public-trust/
• NeurIPS: https://globalgeopolitics.org/ban-on-chinese-scientists-scrapped-at-neurips-after-backlash/
• xAI: https://www.marktechpost.com/2026/04/18/xai-launches-standalone-grok-speech-to-text-and-text-to-speech-apis-targeting-enterprise-voice-developers/
• Microsoft: https://joshbersin.com/2026/04/could-microsoft-win-the-war-for-enterprise-ai/

Welcome to AI RADAR! Today we explore OpenAI's **GPT-Rosalind**, a frontier model set to revolutionize drug discovery and life sciences. We also unpack the White House meeting regarding Anthropic's **Mythos**, an autonomous AI hacking model with severe cybersecurity implications. Plus, OpenAI upgrades **Codex** into a macOS super-app capable of full desktop automation. Finally, we look at **Claude Design** for automated UI prototyping and Perplexity's new **Personal Computer** native app.

Sources:
• GPT-Rosalind: https://venturebeat.com/ai/openai-debuts-gpt-rosalind-a-new-limited-access-model-for-life-sciences-and-broader-codex-plugin-on-github/
• Mythos AI: https://www.pbs.org/newshour/politics/white-house-chief-of-staff-to-meet-with-anthropic-ceo-over-its-new-mythos-ai-model
• Codex Super-App: https://openai.com/index/codex-for-almost-everything/
• Claude Design: https://www.anthropic.com/news/claude-design
• Perplexity: https://www.perplexity.ai/hub/blog/personal-computer-is-here

On this episode of AI RADAR, we explore **OpenAI's GPT-Rosalind**, a revolutionary reasoning model built for life sciences and drug discovery. Next, we unpack the White House's sudden reversal on **Anthropic's "Mythos"**, weighing access to the powerful cybersecurity AI despite previous blacklists. We also break down **Nvidia CEO Jensen Huang's** fiery defense of his hardware dominance against hyperscaler custom chips. Plus, quick hits on **Google Chrome's native AI "Skills"** and tighter search citations in **ChatGPT 5.3**.

Sources:
• VentureBeat: https://venturebeat.com/ai/openai-debuts-gpt-rosalind-a-new-limited-access-model-for-life-sciences-and-broader-codex-plugin-on-github/
• Axios: https://www.axios.com/2026/04/16/trump-officials-anthropic-mythos-access-blacklist
• Business Insider: https://www.businessinsider.com/jensen-huang-comparing-china-chip-sales-nukes-lunacy-2026-4
• Tom's Guide: https://www.tomsguide.com/computing/software/google-just-launched-chrome-skills-and-it-fixes-the-most-annoying-part-of-using-ai
• MarketingProfs: https://www.marketingprofs.com/articles/2026/ai-update-april-17-2026-ai-news-and-views-from-the-past-week