Lecture

Intel CEO Lip-Bu Tan confirmed that Intel will continue to participate in Elon Musk's Terafab chip project.

According to Bloomberg, Tan stated before attending an event in Tokyo commemorating Intel's 50th anniversary in Japan that the company remains committed to collaborating with Musk. This follows earlier reports that Musk was negotiating with TSMC regarding Terafab cooperation, which had dampened market confidence in Intel. Intel signed a contract in April to become a development partner for the project.

Terafab represents Musk's bold attempt to enter advanced chip manufacturing. While global semiconductor and ancillary capacity constraints are driving new production lines, this statement primarily aims to stabilize investor expectations without disclosing new technical milestones or capacity data.

TAPDreamer attack zeroes out world model success rates.

The study proposes an adversarial patch attack against world action models that uses only a public encoder to construct a fixed local perturbation, requiring no access to target model outputs or policy queries.

In closed-loop evaluation, one frozen patch covering about 6.5% of the input reduced FastWAM's success rate from 97.7% to 0.0% across 40 LIBERO tasks and from 90.86% to 0.0% across 50 RoboTwin tasks; matched random patches retained approximately 80% success.

These results are based on author-reported simulated benchmarks and have not yet been independently reproduced or verified on real robots.

WordPress released version 7.1.3 on October 6, fixing seven security vulnerabilities and four bugs in the core platform.

Notably, three of these seven vulnerabilities were reported by the artificial intelligence company Anthropic. The remaining issues were identified by Trail of Bits, Patchstack, independent researchers, and the WordPress security team.

The most exploitable flaw is a stored cross-site scripting (XSS) vulnerability in the comment management interface, which can trigger malicious scripts when an admin opens the page. Anthropic's reported vulnerabilities involve the WXR exporter, denial-of-service risks, and privilege escalation.

OpenAI has released its new Decisions API along with the gpt-6-luna model, which supports image input.

Unlike competitor Jev, gpt-6-luna allows for multimodal data processing. In terms of pricing, OpenAI charges 10 cents per million input tokens and does not charge for output; Jev's equivalent price is 4.2 cents.

This information comes from Simon Willison's blog post dated October 6, 2026, where he built a plugin based on the official documentation.

Google has launched AI Edge Foresight, an experimental macOS app that operates entirely offline.

Powered by EmbeddingGemma 2 and Gemma 4 models, the app accesses system audio and microphone inputs to enhance notes, index conversations, and retrieve personal files without cloud connectivity.

Available as a free download for Apple silicon Macs, it serves as a showcase for on-device multimodal AI capabilities focused on privacy and local productivity.

Anthropic announced on October 6 that it is expanding its Claude Startups program to lower build costs for early-stage AI companies.

New benefits include up to $45,000 in discounts and credits from third-party tools via the "Claude Startup Stack," a one-time $1,000 API credit, and a free year of Claude Team (up to five premium seats).

Eligibility has been broadened to startups founded within the last five years or funded within the last two. Members can also book virtual office hours with Anthropic’s Applied AI team and join community events.

Note that the $45,000 figure is based on partner list prices; actual redeemable value depends on specific offer terms and team size.

Google Docs and Drive now natively support Markdown files.

Previously, users had to convert .md files to Docs format, often losing formatting or comments. Now, files can be opened, edited, and collaborated on directly in Docs, while Drive renders previews including links and tables.

The rollout began October 5, 2026, for all Workspace and personal accounts, completing over up to 15 days.

SignSplit announced on October 5 that it has emerged from stealth with a $400 million strategic seed round led by W Group, valuing the company at $1 billion.

Founded in 2024, SignSplit builds infrastructure for "signed data," enabling individuals and institutions to license their data, work, and likeness to AI systems while retaining ownership and receiving compensation. The platform offers data pools, verification layers, and licensing management.

W Group Founder Volodymyr Nosov stated the investment aims to accelerate global rollout, asserting that industries relying on human data will need such infrastructure. No independent third-party verification of technical deployment scale or user payment status is currently available.

Transsion Holdings has submitted its listing application to the Hong Kong Stock Exchange, planning to issue 86.6 million shares at a maximum price of HK$38.80 per share, with trading expected to begin on October 15.

The offering aims to raise up to HK$3.36 billion (approx. CNY 2.874 billion), primarily allocated to artificial intelligence research and development, including AI assistants and intelligent agents, to enhance product differentiation. Cornerstone investors, including Singapore's Government Investment Corporation (GIC) and E Fund Management, have agreed to subscribe to approximately 37% of the base offering size.

Headquartered in Shenzhen, Transsion owns brands like Tecno and Infinix and holds a 53% market share in African smartphones. Despite its dominance in emerging markets, the company faces margin pressure from rising memory chip costs, and its A-shares have fallen 19% year-to-date. This IPO serves as a litmus test for investor appetite toward companies outside the direct AI supply chain.

SpaceX seeks $40B for Nvidia chips, Economic Observer says.

Per the Financial Times, Apollo would lead the raise. It would include about $10B in loans and $30B in investment-grade bonds, closing in 2027.

SpaceX issued $25B in bonds after its June listing, but debt concerns led to a sell-off. The plan is not confirmed. Relevant parties declined to comment.

Google has been ordered to halt construction on two data centers in Finland.

The Finnish Licensing and Supervision Agency (LVV) issued a notice to Google subsidiary Tuike Finland Oy on Tuesday, suspending building work at projects planned for Muhos and Kajaani until environmental impact assessments are completed. The suspension must be in effect no later than October 23.

The order specifically covers tree removal, topsoil stripping, excavation, quarrying, and road construction. A Google spokesperson stated they understand the concerns and will follow LVV guidance. This comes just one month after Google announced a $15 billion investment in Finnish AI infrastructure, its largest single commitment in Europe.

U.S. Representative Lori Trahan introduced the "CLAIM Act" draft on October 7 to clarify legal liability for AI developers when their agents cause harm.

The proposal directly responds to the rogue AI hack at Hugging Face earlier this year. It aims to make it easier for third parties to sue AI developers while clarifying questions around intent related to AI agents’ conduct.

Unlike efforts to create a single national standard, the bill sets a federal floor without preempting state laws. This allows states to maintain stricter regulations, creating a multi-layered legal environment.

This is currently only a draft bill and has not entered formal voting procedures. Its passage depends on subsequent congressional review and bipartisan negotiations.