AI Security
Security implications of AI, including prompt injection, model abuse, cyber capabilities, data leakage, agent security, and defensive AI.
The First 36 Hours of a Bioweapon Outbreak Are an Intelligence Crisis
Annie Jacobsen argues that a biological release from a Russian high-containment laboratory could present U.S. officials with a narrow 12-to-36-hour containment window precisely when evidence is most uncertain and Moscow may have the strongest incentive to deny what happened. Using a hypothetical explosion at the Vector laboratory near Novosibirsk, she describes an intelligence effort to infer a concealed outbreak from satellite imagery, intercepted communications, deaths and signs of decontamination. The resulting dilemma, she says, is that a response forceful enough to stop a suspected release could also provoke nuclear escalation.
Kimi K3 Shows Open Models Closing the Frontier Gap
John Coogan argues that Moonshot’s Kimi K3 challenges both the assumption that Chinese open-weight models will remain far behind proprietary frontier systems and the view that greater model efficiency weakens demand for AI infrastructure. In his account, efficiency shifts compute bottlenecks rather than removing them, while capable downloadable weights make cyber controls and the protected returns needed to finance frontier training runs harder to sustain. The article presents a related constraint in media: Netflix is putting generative AI into production workflows, but *The Odyssey*’s debut points to the continuing value of audience trust in a director’s name.
Scientific Agents Need Signed Receipts to Coordinate Across Organizations
Armanas Povilionis of Alithea Bio argues that scientific agents will need an auditable supply chain, not simply more tools or API access, if they are to discover services, commission work and spend across organizational boundaries. His proposed Froglet protocol records those interactions in a signed, hash-linked chain from service description and quote through deal, execution and receipt. The aim, he says, is to make evidence of what was agreed and delivered portable across providers, hosts and payment systems.
AI Security Scanning Shifts the Bottleneck to Verification and Patching
Anthropic’s Eugene Yan argues that frontier models have made vulnerability discovery abundant, shifting the practical security bottleneck to verifying, triaging, routing, and patching findings. Drawing on Mozilla’s jump in Firefox fixes and Anthropic’s scan of more than 1,000 open-source repositories, he proposes a six-step harness built around threat models, isolated sandboxes, independent exploit verification, and human-owned remediation. The aim is not to automate scanning alone, Yan says, but to turn model-generated candidates into reproducible, prioritized fixes.
Thinking Machines Bets on Open Weights and Fine-Tuning Services
Thinking Machines Lab is positioning its first model, Inkling, as an open-weight system built for customer customization rather than as the industry’s benchmark leader, TBPN’s John Coogan argues. Led by former OpenAI technology chief Mira Murati, the company pairs the model with a fine-tuning service that could let customers retain control of the weights while paying Thinking Machines to adapt and operate them. Coogan and Tyler Cosgrove also question how cleanly Inkling can be described as free of distillation, given its reported use of synthetic data from another open-weight model.
Japan Recasts Economic Security as Deterrence Against Coercion
Former Japanese national security adviser Shigeru Kitamura argues that Japan must treat military deterrence, economic resilience, technological protection and alliance coordination as parts of a single strategy against coercion that begins well below the threshold of war. With China’s military expansion and economic leverage reshaping the regional balance, he says Japan should reduce one-sided dependencies while building technological and industrial capabilities that make it strategically indispensable. The US-Japan alliance, in his account, must turn that approach into coordinated action on supply chains, critical technologies, energy and defense capacity.
Apple Sues OpenAI Over Alleged AI Device Trade Secrets
Apple’s lawsuit, California’s challenge to Paramount’s Warner Bros. Discovery acquisition, and a new AI warning letter each concern how institutions respond to technological or market disruption. Apple alleges document theft and improper recruiting as OpenAI develops consumer hardware, while the Paramount case turns on whether Hollywood is a distinct antitrust market or competes for attention across media. The letter, signed by more than 200 researchers and economists, does not prescribe a fixed intervention but calls for research and institutional readiness for possible AI-driven economic change.
Calibrated Benchmarks Expose What Raw Accuracy Misses
Alejandro Vidal, founder of Mindmakers, argues that AI benchmarks should stop treating raw accuracy as a sufficient measure of model ability. Drawing on Item Response Theory from psychometrics, he proposes calibrating questions by difficulty and discriminative value, then estimating models on a shared scale rather than simply counting correct answers. Vidal says the approach can expose flawed items, reduce routine evaluation sets while preserving rankings, and flag response patterns consistent with benchmark contamination or operational errors.
OpenAI Takes GPT-5.6 Global After U.S.-Reviewed Limited Preview
Bloomberg’s Seth Fiegerman reports that OpenAI is moving GPT-5.6 from a limited, government-shaped preview to global availability after pressure from the Trump administration to stagger the release. The model family is aimed at coding, cybersecurity and enterprise use, but Fiegerman says it remains unclear what changed between the restricted rollout and the wider launch. OpenAI argues the review process should not become the default for frontier-model releases, even as Washington scrutinizes systems with stronger cyber capabilities.
AI Agent Security Is Becoming an Access-Control Problem
Devvret Rishi, Rubrik’s GM of AI and former Predibase chief executive, argues that the main enterprise AI security risk is not the model but the access granted to agents that can act inside databases, email, code repositories and other systems of record. In a conversation with Craig Smith, Rishi presents Rubrik Agent Cloud as a governance layer for that access: a platform meant to discover agents across an enterprise, monitor their prompts and tool calls, enforce natural-language policies at runtime, and help recover from destructive actions.
Enterprise AI Buyers Are Turning Sovereignty Into a Vendor-Control Fight
The Palantir-Nvidia partnership is presented as evidence that enterprise AI safety is becoming a question of customer control rather than model access. David Sacks, Chamath Palihapitiya, David Friedberg and Jason Calacanis argue that companies and governments should not hand proprietary data, model weights, compute decisions and operating know-how to frontier labs that may later compete with them. The discussion extends from that AI sovereignty argument into a separate jobs dispute over whether current employment data can answer future displacement claims, and into fights over birthright citizenship and California’s budget as questions of institutional authority and fiscal accountability.
AI’s Scarce Inputs Are Rewriting the Open Versus Closed Model Debate
John Coogan, Jordi Hays and Tyler use Zhipu AI’s open-weight GLM-5.2 release to argue that the open-versus-closed AI fight is now about timing, not a settled winner. Closed labs may still lead at the frontier, but they say capable open models arriving close behind can weaken API-based controls, shorten monetization windows and complicate security planning. The discussion broadens that pressure into the AI supply chain, where scarce compute and memory capacity may be capturing profits before model providers can.
Fable and Mythos Recall Targets the Wrong AI Cyber Risk
Alex Stamos, the former Facebook chief security officer and current Corridor chief product officer, argues that the U.S. government’s forced pullback of Anthropic’s Fable and Mythos models misidentified the real cybersecurity threshold in AI. In his account, the decisive shift came earlier with models such as Opus 4 and GPT-5, which made elite vulnerability discovery scalable, while Fable’s risks were not meaningfully distinct from capabilities already available in other U.S. and Chinese models. Stamos says policy should target exploit creation and offensive operations, not bug-finding itself, or risk weakening defenders and making U.S. AI less reliable.
Creator Businesses Are Sorting by Cost, Distribution, and Control
John Coogan and Jordi Hays argue on Diet TBPN that several hyped technology markets are entering a more exacting phase in which distribution, margins and access matter more than slogans. Their discussion frames the creator economy as a sorting of business models, Meta’s smartglasses as a consumer hardware category gaining traction despite weak investor credit, and OpenAI’s limited GPT-5.6 rollout as evidence that frontier AI is now constrained by security policy, infrastructure and control over who gets to use it.
Pandemic Preparedness Is Being Built and Dismantled at Once
At Aspen Ideas: Health, Blythe Adamson, Ashish Jha, John Nkengasong and Jessica Malaty Rivera argued that the world is still not ready for the next fast-moving infectious threat, even as WHO members negotiate the first legally binding pandemic agreement. Their case was that preparedness now depends as much on early detection, engineered-biology safeguards, regional response capacity, trusted communication and global financing as on vaccines and stockpiles — and that several of those systems are being weakened at the same time they are supposed to be built.
The Cloud Is Being Rebuilt Around Agents, Tokens, and Sandboxes
Vercel chief executive Guillermo Rauch used a Stanford MS&E435 seminar to argue that coding agents are expanding the software market rather than merely making developers faster. In his view, AI is widening software creation from professional programmers to business users and autonomous agents, while shifting cloud demand from websites and applications toward deployed agents that need token routing, sandboxes, security, observability and long-running compute.
IBM Bets AI Value Will Come From Orchestration, Not Model Size
IBM chief executive Arvind Krishna argues that the next phase of enterprise AI will be defined less by ever-larger foundation models than by using the right model and infrastructure for each task. Speaking with Masters of Scale host Bob Safian, Krishna says model switching will become easier, AI compute costs are likely to rise, and companies should move from pilots to scaled use cases while questioning the economics of the broader AI buildout. He also frames IBM’s $10bn quantum push as a bid to get ahead of the next hard technology curve.
AI Engineering Is Moving From Model Benchmarks to Production Harnesses
Shawn “swyx” Wang argues that AI engineering is shifting from a race over raw model capability to the production systems that make models usable: evals, harnesses, memory, routing, infrastructure and auditability. Drawing on Cognition’s Frontier Code benchmark and his view of the AI Engineer agenda, Wang says the key software frontier is no longer whether agents can pass tests, but whether they can produce maintainable, mergeable code inside real organizations. His broader case is that unstable model access and enterprise constraints make the surrounding system, not the model alone, the durable product boundary.
SpaceX, Anthropic, and Iran Test the Case Against Centralized Power
The All-In panel uses a week of fights over welfare, SpaceX, Anthropic and Iran to argue over who should hold power when risk is high: markets and individuals, or political and corporate gatekeepers. David Friedberg, David Sacks and Chamath Palihapitiya cast much of the discussion as a warning against centralization, from benefit systems that can weaken agency to AI safety regimes that could hand control to governments and hyperscalers. Jason Calacanis shares parts of that concern but presses the practical tensions, especially in the Anthropic dispute and in Trump’s Iran memorandum, where he questions whether the war that produced a possible deal was necessary.
Export Controls Turn Frontier AI Access Into a Political Problem
John Coogan framed Anthropic’s Fable/Mythos suspension as both an export-control crisis and a sign that frontier AI companies are poorly aligned with Washington’s current political and security instincts. On Diet TBPN, Coogan and Jordi Hays argued that the same access problem is appearing across tech and media: foreign-national limits complicate AI development and sales, Meta’s AI use is being pulled back into budget discipline, and Fox’s reported Roku deal is a bet that control of connected-TV distribution will matter as ad-supported streaming grows.
GRU Space Plans Lunar-Regolith Bricks as the First Step Toward a Moon Hotel
On This Week in Startups, GRU Space founder Skyler Chan argues that a Moon hotel is the first commercial wedge for a larger off-Earth manufacturing business: using lunar regolith to make construction materials rather than shipping them from Earth. Chan lays out a plan to prove the technology by making a brick on the Moon, then scale toward robotic habitats, NASA construction work, space tourism and eventual claims on lunar resources. The same episode turns to Anthropic’s forced shutdown of Fable 5 and Mythos 5, which Jason Calacanis and Lon Harris frame as a warning that frontier capabilities can be cut off before law, politics and operating norms have settled.
Russia-Based Network Used Telegram Recruits to Target Starmer Properties
Financial Times reporter Miles Johnson traces the arson attacks on properties linked to Keir Starmer to a Russia-based online network that allegedly recruited a 21-year-old Ukrainian in London through Telegram. Johnson’s account argues that Roman Lavrynovych was moved from posting far-right propaganda to vandalism and then fire-setting without being told the political significance of the targets. The case is presented as an example of Russian-linked disruption that is cheap, deniable and designed to look like local extremism.
Codex Adds Chrome DevTools Access for Web App Debugging
OpenAI says Codex’s Browser Use can now connect to the Chrome DevTools Protocol, allowing it to inspect running web applications through console logs, runtime errors, local storage, styling, network traffic and performance profiles. The source argues that this moves Codex debugging beyond code inspection: in a slow chat-app example, Codex profiles interactions, identifies duplicate requests and expensive server paths, makes targeted fixes, and reports before-and-after timings. The capability is gated behind Developer mode and per-site approval because CDP access can expose sensitive browser internals.
AI Works Best When Domain Experts Control Its Use
Josh Tyrangiel’s AI for Good argues that artificial intelligence is most useful when domain experts, not technology companies or models themselves, decide how it is applied. In conversation with Aspen Economic Strategy Group director Melissa S. Kearney, Tyrangiel says his reporting found real gains in healthcare, education, government, and recycling, but mostly as incremental improvements shaped by doctors, teachers, public servants, and other practitioners. His case is not that AI’s risks are overstated, but that the policy question is how to preserve human authority while regulating the most dangerous capabilities.