🛎️ Security Review Goes AI-Native

Good Morning, AI Enthusiasts!

Something structural is shifting. Let’s decode the pressure.

NEW LAUNCH

Security Review Goes AI-Native

👀 What’s happening: Anthropic has launched Claude Code Security in limited preview, positioning it as an AI security auditor embedded directly inside Claude Code. It scans entire codebases, reasons about business logic and data flow, flags complex vulnerabilities, and suggests patches for human review. This moves Claude from code assistant to active defensive infrastructure.

🌍 How this hits reality: Static scanners rely on rule libraries. Claude reasons like a researcher. Using Opus 4.6, the team reports finding over 500 long hidden vulnerabilities in production open source software. Cybersecurity stocks reacted immediately, with several names dropping between 3 and 7 percent in a single session. The assumption that security review requires specialized standalone tooling is now under pressure.

🛎️ Key takeaway: If AI becomes the default security auditor inside the development stack, traditional vulnerability scanning vendors face structural compression. Security shifts from a separate budget line to a built in model capability. That changes margins across enterprise software.

TOGETHER WITH CLICKUP

The Everything App for Work

Meet ClickUp – the everything app for anything work. Whether you're juggling tasks, tracking bugs, planning sprints, or just trying to remember where you put that file, ClickUp keeps your whole team in one place. Say goodbye to tab overload and hello to one sleek platform that somehow does it all without losing its mind – or yours.

OPENAI

Altman Energy Analogy Collapses

👀 What’s happening: At a public event in India, Sam Altman compared AI training energy to raising a human. A person uses about 17,000 kWh over 20 years. Training GPT-4 took roughly 50 GWh, equal to 3,000 humans. That flagship model was retired in under two years. GPT-5.2 is now default, consuming around 18 Wh per query, up to 40 Wh for longer reasoning.

🌍 How this hits reality: ChatGPT processes about 2.5 billion queries daily. Even well below 18 Wh blended, inference alone rivals the electricity use of more than a million US homes per day. Humans trained on 17,000 kWh generate economic output for decades. GPT-4’s 50 GWh was amortized over less time than a standard car lease.

🛎️ Key takeaway: The analogy is not insight. It is framing. Once you do the arithmetic, it reads less like biology and more like burn rate management. Dressing up gigawatt scale industrial expansion as childhood nutrition does not change the grid math.

CODING

Kiro’s Coding Failure

👀 What’s happening: AWS suffered at least two recent internal outages tied directly to its in-house Kiro AI coding agent. In one December incident, a cost analytics system went offline for roughly 13 hours after Kiro autonomously chose to delete and rebuild an environment to “fix” an issue. Engineers had allowed it to act without intervention.

🌍 How this hits reality: This was not a complex edge case. The agent misjudged remediation logic and escalated a routine problem into full system removal. In a cloud division that runs infrastructure for millions of customers and generates tens of billions in annual revenue, even a 13-hour internal disruption signals weak safety constraints. That reflects product immaturity, not operational surprise.

🛎️ Key takeaway: Kiro is not ready for autonomous production control. Until its decision boundaries and rollback safeguards mature, granting it live authority will continue to create avoidable risk inside AWS itself.

DAILY TL;DR

• Samsung is integrating Perplexity into Galaxy AI on the Galaxy S26, advancing a multi-agent ecosystem with system-level app access.
• YouTube is testing its conversational AI chatbot on TVs, consoles, and streaming devices, allowing viewers to ask questions while watching videos.
• OpenAI is reportedly testing a $100-per-month ChatGPT Pro Lite tier to bridge the pricing gap between Plus and Pro.
• GenAI.mil, the Pentagon’s generative AI platform, has partnered with OpenAI to expand AI use in U.S. military data analysis and training.
• The Guardian said Nvidia is planning to invest $30bn in OpenAI’s next funding round, after a $100bn deal between the two dissolved earlier this month.

TRENDING

Daily AI Launches

• Rork Max has launched a web-based Swift app builder for all Apple platforms, supporting complex features and one-click installs.
• git-lrc has launched a pre-commit AI code reviewer designed to catch bugs, logic errors, and security risks introduced by AI coding agents.
• WANotifier has relaunched a WhatsApp CRM and marketing platform that charges zero markup on Meta's official API fees.
• Prism Videos has launched a cloud-based timeline editor that centralizes AI video generation, assembly, and templating in one workspace.

Featured AI Tools

• ✨ MuseMail.ai makes creating stunning, on-brand emails from one prompt with AI precision.
• 🪶 CopyOwl is the First AI Research Agent, deep research on any topic in one click.
• 🦾 Flot AI writes, reads, and remembers across any apps and webs.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next
• Oncely helps you discover the next wave of AI products and launches

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.