The AI Sec-Pocalypse Is Actually Nigh?

Key takeaways

AI is increasingly being used to discover and exploit zero-day vulnerabilities, leading to a cybersecurity landscape where traditional patching policies are no longer sufficient.
The 90-day vulnerability disclosure window is effectively dead in an era where exploits can be developed and weaponized within minutes, necessitating immediate patching and AI-driven security responses.
OpenAI is expanding its enterprise presence through a major investment and acquisition, aiming to embed AI deployment capabilities directly within organizations.
Apple continues to refine its software, focusing on stability and AI features, with major updates expected at WWDC 2026.
TikTok’s new ad-free tier in the UK reflects evolving user preferences and regulatory influences, offering greater control over ad exposure.
The future of AI infrastructure will shift from speed-focused architectures to memory-focused designs, enabling autonomous, agentic AI operations that do not rely on human-in-the-loop processes.

Summary

The Emerging Threat of AI-powered Cyberattacks

Google recently reported the first known incident of hackers using AI to discover and weaponize a zero-day vulnerability. The attack was sophisticated, relying on AI to identify vulnerabilities faster than traditional methods. Experts warn this is just the beginning, and the proliferation of AI in cyberattack tools could lead to more frequent, faster exploits that traditional patching timelines can't keep up with. The 90-day vulnerability disclosure policy is considered obsolete, as AI enables hackers to develop and deploy exploits in mere minutes, putting enterprise security at an unprecedented risk.

Revolution in Software Security and Patch Management

The rapid development of AI tools has fundamentally altered how vulnerabilities are exploited and patched. Skilled reverse engineering may soon be replaced by AI, drastically reducing the time needed to convert patches into exploits. This shift makes traditional safety nets, like scheduled patch cycles, ineffective, as attackers can respond instantly. Experts advocate for integrating AI into security workflows, including at the code deployment stage, to stay ahead of automated exploits.

OpenAI’s New Deployment Venture and Industry Push

OpenAI has announced the creation of a new company with over $4 billion in funding to assist organizations in deploying AI. This venture, which includes acquiring AI consulting firm Tomorrow, aims to embed AI expertise directly into enterprises, helping them integrate AI systems effectively. As AI adoption in the enterprise accelerates, OpenAI aims to maintain its competitive edge, focusing on scaling solutions and fostering large-scale deployments across various industries.

Refinements in Apple’s Software and Hardware Plans

Apple plans to refine macOS 27 to address issues related to Liquid Glass, with a focus on improving user interface and stability. Additionally, new versions of iOS and iPadOS are expected to introduce interface tweaks and more AI features, including a redesigned Siri that acts as a proactive assistant. These updates aim to enhance the user experience through polishing and AI integration, announced to debut at the Worldwide Developers Conference.

TikTok’s Ad-Free Subscription in the UK

TikTok is rolling out an ad-free tier in the UK, costing about £3.99 per month for users aged 18 and over. This subscription eliminates ads and reduces data collection for advertising purposes, purportedly in response to GDPR regulations. The move aims to offer users greater control over their experience while supporting British businesses that leverage the platform’s advertising tools.

The Future of AI and Compute Infrastructure with Agentic Inference

Ben Thompson explores how AI's evolution from answer inference to agentic inference will revolutionize compute infrastructure. Unlike today's models, future AI agents will operate independently, performing tasks without human involvement, demanding new memory architectures focused on capacity rather than speed. This shift implies a fundamental change in system design, where storage and memory hierarchies become more important than raw compute power, enabling AI to work autonomously at massive scales.

Show notes

Google reported the first known case of hackers using AI to discover and weaponize a zero-day vulnerability. OpenAI launched a $4B+ deployment company and acquired Tomoro. Apple plans Liquid Glass refinements for macOS 27, TikTok rolls out an ad-free tier in the UK, and Ben Thompson argues agentic inference will reshape compute.

Google's TIG reports the first known example of hackers using AI to discover and weaponize a zero-day; TIG's chief analyst says "this is the tip of the iceberg" (NYT)

The 90-day vulnerability disclosure policy is dead, as LLMs compress bug finding and exploit development time, and critical issues must be patched immediately (Himanshu Anand)

OpenAI launches the OpenAI Deployment Company with a $4B+ investment to help organizations build and deploy AI systems, and acquires AI consulting firm Tomoro (Reuters)

Sources: Apple is working on a "slight redesign" for macOS 27 to address Liquid Glass issues and plans a feature to automatically group Safari tabs in "27" OSes (Bloomberg)

TikTok is rolling out TikTok Ad-Free, a £3.99-per-month subscription for UK accounts aged 18 or older "over the coming months", after testing the option in 2023 (TechCrunch)

Agentic inference is set to be different than today's inference, and will change compute infrastructure because speed won't matter when humans aren't involved (Stratechery)