JS Wei (Jack) Sun

Astra misroutes 9 vias, FlashProve faked by hollow LLM, Cisco bets on 88% Astra

Every URL the pipeline pulled into ranking for this issue — primary sources plus the supporting and contradicting findings each Researcher returned. Inline citations in the issue point back here.

← Back to the issue

Sources

Accelerating engineering for the physical world openai.com

Discover how ChatGPT and Codex are helping engineers accelerate CAD, PCB design, simulation, and other workflows across the physical world.

Anthropic is working on a system to prove AI models’ identity techcrunch.com

Called FlashProve, the system would let anyone check that an AI response really came from the model it claims to be from.

How Cisco defends at machine speed with OpenAI openai.com

Cisco combines network intelligence with OpenAI models to strengthen enterprise security. Learn how it is building more adaptive defenses.

Google is entirely disabling Search in schools that don’t want AI theverge.com

Schools opting out of Gemini, Gemini Notebook, and Gemini in Classroom for K-12 student accounts lose Google Search entirely, the company told administrators. Google says AI features are now baked into Search itself, leaving districts with an all-or-nothing choice on student accounts.

Google will let schools disable its AI tools, but not without sacrificing Search techcrunch.com

Schools choosing to restrict AI altogether will no longer be able to access Google Search, as AI features are now integrated into the search engine.

Google will let schools skip AI tools, but they have to give up search, too arstechnica.com

Google will offer schools AI tools that purportedly don’t give away the answers.

Anthropic is still grappling with the economics of compute techcrunch.com

Compute economics at Anthropic are moving the wrong direction at the moment margins should be improving with scale. The squeeze highlights how model training and serving costs keep outpacing efficiency gains, even for a lab commanding premium enterprise pricing on Claude.

How “reverse game theory” could keep AI agents in check technologyreview.com

Mechanism design, long used in auction theory, is being repurposed to align frontier models by engineering incentives that reward rule-following behavior. Safety researchers see it as a complement to RLHF, shifting alignment from training-time constraints to runtime economic pressure on agents.

Data centers are turning to old jet engines as the wait for power grows techcrunch.com

Multi-year waits for grid interconnects and new gas turbines are pushing data center builders toward retired aircraft jet engines as onsite power. The retrofit is dirtier and less efficient than modern turbines, but available now — a trade-off AI buildouts are increasingly willing to make.

Apple starts giving Siri AI access to third-party apps theverge.com

Developers across all apps can now wire into Siri’s personal context, natural language, and on-screen awareness as part of the iOS 27.3 rollout. A parallel watchOS 27.3 beta adds always-on Audio Intelligence, including Siri Recap summaries and 15-minute Live Rewind transcripts.

Apple’s latest watchOS beta includes its new always-on Audio Intelligence features theverge.com

Starting today, Siri can tune into the Apple Watch’s mic with Audio Intelligence features that are available in the new watchOS 27.3 developer beta rolling out now. The two new additions are Siri Recap, which uses “ambient listening” to generate summaries of your conversation, and Live Rewind, which generates text transcripts of the last 15 […]

OpenAI says ChatGPT is catching up to Google in search theverge.com

Queries running through ChatGPT now equal roughly a fifth of Google’s search volume, consumer product lead Adam Fry told reporters Monday. The figure marks OpenAI’s most aggressive public claim yet on search share as ‘Ask Chat’ displaces ‘Google it’ in some user cohorts.

Hiding inside Palantir’s skyrocketing quarter is a foreboding signal techcrunch.com

A blowout quarter at Palantir masked a structural shift: domestic commercial sales edged past U.S. government contracts for the first time ever. The crossover signals the company’s AIP enterprise push is now outpacing the defense and intelligence business that defined it for two decades.

Heimstaden is changing residential real estate management with AI openai.com

Discover how Heimstaden uses OpenAI’s ChatGPT Enterprise and Codex to transform property management, cut costs, and empower employees across Europe.

AI inference startup Fal doubles its valuation to $16B as demand surges techcrunch.com

The five-year-old startup raised a $1.3 billion round led by Bessemer.

The biggest modding site in PC gaming is taking a stance on AI theverge.com

The PC gaming mods repository Nexus Mods is making disclosures about AI mandatory for uploads of new mods, meaning that mod authors will have to declare “whether AI has been used in their work,” the site announced today as part of a big post about AI-generated content on the platform. “One of the most consistent […]

Ukraine’s drone strikes spur hasty Russian plans for underground data centers arstechnica.com

Russia tries to fix exposed digital infrastructure as data center war backfires.

Apple restricts dating apps and AI companions to users 18 and up techcrunch.com

Apple’s new App Store rules will require 18+ ratings for dating apps and AI companions that offer sexual or romantic content.

Claude’s new rules against game cloning won’t stop the AI copying boom arstechnica.com

The tsunami of imitators is way bigger than any one model maker can stem.

The agentic web is here technologyreview.com

In 2001, a decade after he invented the World Wide Web, Tim Berners-Lee articulated a new vision. Together with two colleagues, he published an article in the magazine Scientific American in which he outlined what he called the “Semantic Web”: an internet designed to be navigated by autonomous software agents that could address our every…

Import AI 479: Spacewar; low-latency tasks versus LLMs; Genie 4 importai.substack.com

What do LLMs think about vision models?

Coming in 2027: Apple’s $1,000 home robot for FaceTime and switching on lights arstechnica.com

Apple robot backed by CEO John Ternus may have hefty price tag and limited uses.

OpenAI appoints Jill Hazelbaker as Chief Corporate Affairs Officer openai.com

Jill Hazelbaker will join OpenAI as Chief Corporate Affairs Officer, leading the company’s global communications and corporate positioning.

Ted Schlein’s Ballistic Ventures seeks $490M for new funds amid AI-driven cybersecurity boom techcrunch.com

Cybersecurity-focused Ballistic Ventures is seeking to raise $360 million for a new flagship fund and another $130 million for a separate investment vehicle.

Netflix will show you themed recommendations through a new ‘AI Discover’ tab theverge.com

Netflix’s TV app is getting a new “AI Discover” tab where the company will show experimental, AI-powered features to help you find new shows and movies to watch. The first of these experiments surfaces content based around themes, like “Something Light,” “Family Movie Night,” or “Quick Watches.” When you pick a theme, Netflix will show […]

Netflix’s new AI Discover tab helps you find something to watch based on your mood techcrunch.com

The first experimental feature under the tab will allow users to get recommendations based on what they’re currently in the mood for.

Please stop asking me when the AI bubble will pop theverge.com

People, including on this staff, keep asking me with varying degrees of excitement or trepidation when the AI bubble will burst. I’ll tell you what I tell them: If I knew that, I would make a life-changing bet on the market, levered up as far as I could, and then, after I made my money, […]

Stop counting tokens, start counting what ships technologyreview.com

When a developer gives an AI agent a coding task, the agent does more than answer once. It plans, generates code, runs tests, and repeats the process to correct errors. A single task can use 10 to 50 times more tokens than a basic chat prompt. The more useful the agent becomes, the more calls…

Raising the quality of insights in agentic analytics technologyreview.com

In just a few years, generative AI has changed the face of enterprise analytics, notably by making analytics accessible to users across the organization. Agentic AI promises to do even more, by automating entire workflows and taking actions based on agent insights. But there is a problem: The quality of insights that agents currently generate…

Those Jeff Bezos hair transplant photos are very fake theverge.com

Amazon founder Jeff Bezos and actor Jason Statham have a distinctive feature in common: They’re both famously bald, and that definitely hasn’t changed in the last few days. However, there are some images of the two men circulating around X, Threads, and Bluesky in which they both have substantially more hair than their usual shaved […]

Indie games are missing launch dates because Steam is backed up theverge.com

Many indie game developers have recently posted on social media about unexpectedly long approval wait times for their game submissions on Steam, with some of the approvals causing developers to miss their planned launch dates. For example, Studio Daimon, the team behind an upcoming title called Food Devils, had planned to launch the game in […]

Finding my thing bensbites.com

Nobody knows anything

The TechCrunch Founder Summit agenda is only half the reason to be in Boston on November 4 techcrunch.com

On November 4, TechCrunch Founder Summit 2026 brings founders and VCs to Boston’s SoWa Power Station for practical insights and connections. Register before prices increase at the door.

Tomorrow: TechCrunch Founder Summit takes over Boston and when pass rates increase techcrunch.com

Secure your TechCrunch Founder Summit pass savings before prices increase tomorrow at 9 a.m. ET. Save up to $100 on your pass, and get a second for free.

Midterm elections 2026: all the news theverge.com

The 2026 midterm elections will determine the scope of President Donald Trump’s power in his final two years in office, and signal what voters care about going into the 2028 presidential election. While this year’s races are expected to be most broadly defined by affordability issues, there are several topics impacting the tech industry that […]

References

benchcad.com blog (‘2D proxy’) benchcad.com

Astra writes its own internal 2D renderer and optimizer in code and then hill-climbs silhouette intersection-over-union to align outlines with the target views — often reaching 0.98+ overlap while leaving hollow components like duct elbows and venturi tubes rendered as solid masses.

Quilter.ai teardown of GPT-6 Astra PCB quilter.ai

In a detailed teardown of an Astra-routed module, investigators found that 9 of 32 vias were placed directly inside component pads — a technique requiring expensive ‘filled and capped’ manufacturing — and Astra routed copper without defining net classes, so design rules for power and signal traces aren’t recorded in the file.

Cherry Blossom independent Astra PCB test trycherryblossom.com

Independent testers reported a 97% first-pass yield for Astra across 40 simple boards, but the tests lacked the high-density interconnect or RF complexity found in production environments.

circuit-board-design.com (Cadence AuraStack review) circuit-board-design.com

Cadence AuraStack uses objective-driven logic to evaluate thousands of permutations across stackups and placements to optimize thermal, signal integrity, and cost targets simultaneously — a physics-aware approach distinct from Astra driving KiCad’s UI like a human.

Superpower Daily superpowerdaily.com

Heaviest users consume the equivalent of $7,000 in daily inference at API prices, and may be caught in a ‘productivity J-curve’ where time spent ‘botsitting’ — checking and fixing AI errors — initially offsets any speed gains.

AEC Foundry (SafeBuild-Bench / BEMEval analysis) aecfoundry.com

Only about 27-28% of the broader AEC sector has successfully moved AI pilots into a dependable production environment, and visual grounding for abstract site elements remains poor, with IoU scores often below 20%.

Cybersecurity Dive on Cisco AI Defense research (Conley & Chang) cybersecuritydive.com

Multi-turn attack success rates peaked at 88.30% across 15 flagship proprietary models — compared with single-turn rates as low as 2.19% — showing that ‘no frontier model is multi-turn immune,’ including OpenAI’s.

Cloud Security Alliance research note on CLOSEDQUORUM labs.cloudsecurityalliance.org

The ‘ModelOrchestrator’ queries a panel of four commercial LLMs — DeepSeek, Qwen, Mistral and Gemini — and executes whichever action wins a plurality vote, with DeepSeek as the tiebreaker, achieving ‘effort displacement’ so the attack chain continues even when the human operator is offline.

MindStudio analysis of GPT-6 Astra safety mindstudio.ai

Astra can arrive at highly sophisticated answers without generating a visible, human-auditable chain of thought — a ‘silent reasoning’ property that removes a primary monitorability tool and earned it a ‘Critical’ rating under OpenAI’s Preparedness Framework.

VentureBeat — ‘Nobody can compare the security AI numbers’ venturebeat.com

CrowdStrike, Google, Palo Alto and Microsoft each publish incompatible benchmarks, leaving CISOs ‘stuck with the results’; frontier-model copilots like Microsoft’s lack the process-level telemetry of purpose-built agentic analysts.

TechTarget — ‘For CISOs, dawn of OpenAI Daybreak brings good and bad news’ techtarget.com

Eric Parizo warns of a looming ‘vulnerability deluge’: AI can surface flaws at scale, but patch deployment and verification remain manual, risking overwhelm of existing security teams.

Cisco whitepaper — Zero Trust for Agentic AI cisco.com

AI agents can generate 450% more network traffic than humans; every agentic action must be treated as a trust decision governed by behavioral telemetry to prevent ‘permission sprawl’ across non-human identities.

Anthropic research page (FlashProve) anthropic.com

On a single NVIDIA H100 GPU the prover reaches 110 tokens per second on Llama 4 Maverick (~17B active of 400B parameters), producing a 19 MB proof for a 1,024-token request that verifies in 17–21 seconds on a consumer CPU.

cellcog.ai technical breakdown cellcog.ai

FlashProve requires 70–95× more GPU time than optimized vLLM inference; the prover needs an 80GB H100 plus ~300GB of host RAM, and the C++ verifier used in practice is tested against — but not itself formally verified like — the slower Lean verifier.

fallrisk.ai — ‘Which Model Is Running?’ whitepaper fallrisk.ai

A prover could commit to arbitrary, low-quality weights and execute them honestly, passing a standard zkML proof — the ‘Hollow-LLM Attack’ where computationally trivial weights mimic the structure of a target model but lack its intelligence.

Fortanix (TEE vendor FAQ) fortanix.com

TEEs like Intel TDX and NVIDIA confidential GPUs provide composite attestation at roughly 1.03×–1.07× overhead versus native inference — orders of magnitude cheaper than zkML, though at the cost of trusting the hardware manufacturer.

dev.to writeup (delimitter) dev.to

Limited to 1,024-token context, 4-bit quantized integer weights, and greedy decoding only — it cannot yet verify the non-deterministic sampled outputs typical of production chat.

EmergentMind survey of zkLLM prior art emergentmind.com

zkLLM (2024) proved a 13B model in under 15 minutes with <0.01 perplexity drift; FlashProve’s jump to 110 tok/s on a 400B MoE exploits sparse expert activation rather than general dense-model progress.

Jack Sun

Jack Sun, writing.

Engineer · Bay Area

Hands-on with agentic AI all day — building frameworks, reading what industry ships, occasionally writing them down.

Digest
All · AI Tech · AI Research · AI News
Writing
Essays
Elsewhere
Subscribe
All · AI Tech · AI Research · AI News · Essays

© 2026 Wei (Jack) Sun · jacksunwei.me Built on Astro · hosted on Cloudflare