Apple has finally brought Siri AI to the forefront.
1 | Apple Relaunches Siri AI, Device Entry Point Starts Integrating External Models
On June 8, Apple unveiled Siri AI at WWDC. TechCrunch, The Verge, Reuters, and the Financial Times all see this release as Apple's official return after a two-year AI delay. The new Siri is no longer just a voice assistant; it can access personal device data, understand camera scenes, process photos, search emails, and expand into macOS, watchOS, and Home cameras. CNBC and Wired also mentioned that Apple's state-of-the-art AI capabilities will be on par with Google and NVIDIA.
This is not a story of Apple catching up to ChatGPT, but rather Apple acknowledging that the device entry point needs external computing power and external models to fill the gap. Apple's strengths still lie in its ecosystem, privacy, and default apps. The weakness is the speed of cutting-edge models. It is now connecting the two. On the surface, it is a Siri upgrade; at its core, Apple, while retaining the iPhone entry point, is sharing the AI supply chain with Google, NVIDIA, and its private cloud. (Continued from the June 5 report)
(Source: TechCrunch / The Verge / Reuters / Financial Times / CNBC / Wired)
2 | OpenAI Submits S-1, Leading AI Company Enters Public Market Race
On June 8, OpenAI confirmed that it has submitted a confidential S-1 to the U.S. SEC. Reuters, AP, Axios, and The New York Times have all placed it after Anthropic for interpretation. Anthropic submitted its documents on June 1, with OpenAI following a week later. The Verge reported that OpenAI has yet to decide on the IPO timing; some matters are easier to address in private, but the filing has put it on the public market timeline.
What truly changes this trajectory is the capital discipline of AI companies. In the past, cutting-edge labs relied mainly on private valuations, strategic investments, and cloud provider contracts to sustain their expansion. After going public, they will face more transparent risk disclosures, financial metrics, and secondary market fluctuations. While model capabilities remain the narrative core, what investors will look at will shift to computing costs, revenue quality, gross margins, and governance structure. AI companies are not transitioning from the lab to the stock market but from narrative financing to accountable financing.
(Source: Reuters / AP / Axios / NYT / The Verge)
3｜Microsoft Open Source Packages Injected with Data Theft Code, AI Agent Expands Supply Chain Attack Surface
Ars Technica reports that multiple Microsoft open source packages have once again been injected with credential-stealing code. Researchers stated that 73 crypto verification packages were automatically flagged as malicious by GitHub's system. The attack vector is specific: when developers use AI coding agents to open these packages, the malicious code steals passwords and sensitive credentials. Additional analysis from TechCrunch and StepSecurity also points to the same issue, indicating a shift from package manager poisoning to AI agent workflow hijacking.
Traditional supply chain attacks relied on developers installing, running, or copying code. AI programming has automated this layer. The agent will open, read, execute, and patch code on behalf of humans, thereby expanding the attack surface for threat actors. The security perimeter of software is no longer just "can this package be trusted," but rather "which agent, under what permissions, read this package." AI has accelerated development speed but has also turned developers' machines into more easily automated reconnaissance entry points.
(Source: Ars Technica / TechCrunch / StepSecurity)
4｜H-1B Fee Overturned, High-Skilled Talent Mobility Returns to Judicial Boundaries
Bloomberg and The Washington Post report that a U.S. court has overturned the Trump administration's rule imposing a $100,000 fee on the H-1B high-skilled visa. This fee would have significantly increased the cost for tech companies to bring in engineers, researchers, and AI talent. For Silicon Valley and AI firms, the H-1B is not just an ordinary immigration policy but a part of their talent supply chain.
While the news appears to be about a visa fee being blocked by the court, at its core, the competitiveness of the U.S. in AI still depends on global talent flow. The Trump administration was aiming to both take equity and industrial gains from AI companies while trying to raise the cost of foreign high-skilled labor. This court ruling has brought the policy back to procedural and jurisdictional matters. AI industrial policy is not only made up of subsidies and export controls; it also includes who can enter U.S. firms, who can stay in model teams, and who can bring knowledge into the next generation of infrastructure.
(Source: Bloomberg / The Washington Post / Axios)
5｜Agent Accesses Wallets, Research Tools, and Home Cameras, Permission Concerns Arise
An AI agent is stepping out of the chat box. The Block reports that MetaMask has launched the Agent Wallet, allowing AI bots to have Ethereum self-custody access. On the same day, The Verge and Ars Technica mentioned that Google upgraded NotebookLM to Gemini 3.5, incorporating cloud computer and Antigravity support, with research tools taking on more tasks. Apple Home has also integrated Apple Intelligence into cameras, enabling video descriptions generation and supporting natural language search.
Putting these signals together, the issue is no longer whether the agent can answer questions, but what permissions the agent receives. Wallet permissions, file permissions, camera permissions, and browser permissions are essentially all about agency. In the past, applications merely carried out user commands, but now agents are starting to execute commands on behalf of users. The next round of application layer competition will revolve around permission design: who can enable the agent to do more things while reassuring users that it will not overstep its bounds.
(Source: The Block / The Verge / Ars Technica)
Also worth noting ↓
Google and Nvidia are reportedly considering Intel as an alternative chip manufacturer. The Information reports that the two companies are assessing manufacturing redundancy outside of TSMC. The supply chain risk of advanced processes is shifting from geopolitical discussions to concrete procurement options. (Source: The Information)
UK unveils $1.5 billion AI hardware plan. Reuters and Wired report that the plan includes supercomputing and chip funding. Sovereign AI is no longer just about training local models, but about rebuilding hardware, computing power, and local supply chains. (Source: Reuters / Wired)
Goldman, JPMorgan explore trading compute futures. The Information states that Wall Street is exploring using compute futures to hedge AI financing risks. Compute power is transitioning from a cost item to a financial contract, and AI infrastructure is beginning to be unbundled and priced in the capital markets. (Source: The Information)
Meta removes face recognition code from smart glasses app after WIRED report. This is not a formal product launch but a preemptive code-level brake. Privacy concerns for wearable devices are undergoing public and regulatory scrutiny before product launch. (Source: WIRED)
StepStar Plan files for Hong Kong IPO. According to WSJ, this Chinese AI startup is preparing for a Hong Kong stock market listing. While U.S. AI companies rush to NYSE and Nasdaq, Chinese AI firms are exploring different public market exits. (Source: WSJ)
Over 200 crypto organizations, including Coinbase and Ripple, urge the Senate to advance the Clarity Act vote. The crypto industry continues to push regulatory certainty to Congress. Beyond stablecoins, exchanges, custody, and token issuance also need a more defined set of boundaries. (Source: The Block)
Trump threatens tariffs on Brazil over Amazon rainforest deforestation. Fortune reports that the White House is putting the climate issue back into the trade toolbox. Tariffs are expanding from industrial protection to a more flexible political lever. (Source: Fortune)
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
