Daily brief at 7am Melbourne. Unsubscribe any time.

Saturday 11 July 2026

Apple vs OpenAI: The Trade Secret Lawsuit That Could Reshape Silicon Valley's Talent Wars

Apple sues OpenAI for trade secret theft, a vishing campaign hijacks Microsoft 365 via fake passkey enrolment, and Europe's Chat Control 2.0 surveillance law finally passes — here's what actually matters today.

Lead story

Apple vs OpenAI: The Trade Secret Lawsuit That Could Reshape Silicon Valley's Talent Wars

Apple has filed suit against OpenAI, alleging the company systematically stole trade secrets — and claims the misconduct was directed by OpenAI's senior leadership, including a long-serving former Apple employee who made the jump to the AI lab. This is not a patent dispute over who invented something first. Apple is alleging deliberate, directed theft of confidential information by people who knew exactly what they were taking.

The timing is awkward for OpenAI on multiple fronts. The company is navigating a leadership vacuum after Fidji Simo stepped down from the No. 2 role following a prolonged medical leave — precisely when OpenAI needs a steady hand as it pursues a possible IPO and battles Anthropic for enterprise market share. A high-profile trade secret lawsuit from the world's most valuable company is not the kind of headline that helps with either.

What Apple is specifically alleging remains thin in public filings at this stage. But the general shape of the claim — insiders crossing over and bringing proprietary knowledge with them — is a pattern Silicon Valley has litigated for decades. The difference here is the parties involved. Apple has one of the most aggressive trade secret enforcement cultures in the tech industry, and OpenAI is currently the hottest company on the planet. This one will drag on, get expensive, and generate a lot of discovery that neither side wants made public.

Why it matters beyond the courtroom. Apple's own AI strategy has been notably cautious — slow to ship, tight on privacy claims, largely on-device. If the lawsuit reveals that OpenAI had inside knowledge of Apple's unreleased AI roadmap or model architecture choices, it would partly explain why Apple's public offerings have felt curiously well-paced relative to what competitors are building. Or it could be a much more mundane IP dispute dressed up in alarming language. That's the thing about trade secret cases filed by Apple's lawyers: they're written to sting regardless of what's underneath.

What to watch for. First, OpenAI's formal response — whether they deny the conduct outright, or quietly indicate a settlement is already in progress. Second, whether the named former employee is identified publicly, which would trigger a wave of reporting on exactly what they worked on at Apple and what they shipped at OpenAI. Third, how this interacts with the Microsoft-OpenAI relationship: GPT-5.6 just landed in Microsoft 365 Copilot with considerable fanfare, and Apple has its own deeply intertwined AI partnership infrastructure to manage.

For Australian observers, this is worth watching through a different lens. The ACCC has been sharpening its focus on AI market concentration, and a lawsuit alleging that one AI giant's capabilities were seeded by stolen knowledge from a hardware giant goes directly to questions of competitive fairness in a market where Australia has essentially no domestic players at this scale. The outcome won't change what's available on the App Store, but it could meaningfully reshape the liability landscape around AI talent mobility globally — including here.

Also today

Progress Software Tells ShareFile Customers: Shut It Down Now

Progress Software has issued an urgent directive to ShareFile customers running on-premises Storage Zone Controllers, telling them to take the Windows servers offline immediately. The company confirmed it is responding to a "credible external security threat" and has proactively disabled access to affected accounts. Progress hasn't disclosed what the vulnerability is or whether it's being actively exploited, which is a red flag in itself — the company learned harsh lessons from the MOVEit breach in 2023 and appears to be moving faster this time. ShareFile is widely used in legal, financial services, and healthcare sectors. Australian organisations using ShareFile's managed file transfer capabilities should confirm their deployment model with their vendors immediately.

The Hacker News

Vishing Attacks Are Hijacking Microsoft 365 Accounts Via Fake Passkey Enrolment

A threat actor tracked as O-UNC-066 is running a sophisticated voice phishing campaign against Microsoft 365 users. Victims receive a phone call impersonating a security alert, then get directed to a fake Microsoft Entra ID page that captures their credentials and tricks them into enrolling the attacker's passkey — effectively handing over persistent, phishing-resistant access. Okta researchers who discovered the campaign note it's targeting organisations across multiple sectors and is explicitly designed to circumvent MFA. The passkey enrolment angle is clever: it turns a next-gen authentication method into an access vector by inserting the attack at the moment of enrolment rather than authentication. M365 admins should audit recent passkey enrolments and lock down who can register new credentials.

The Hacker News

Europe Passes Chat Control 2.0 — Mandatory Message Scanning Is Now Law

The European Parliament has passed Chat Control 2.0, requiring companies including Google, Meta, and Microsoft to scan users' private messages for child sexual abuse material. Privacy advocates have fought the law for years, arguing that any system capable of scanning encrypted messages fundamentally breaks end-to-end encryption — you can't build a backdoor only good guys can use. Supporters argue the status quo lets CSAM circulate freely behind the shield of encryption. The law now applies to platforms operating in the EU, meaning Australian users of those platforms may see policy changes as companies try to build globally consistent implementations. Expect legal challenges immediately, and watch how Apple — already in a fight with OpenAI — responds given its strong privacy positioning.

The Record

HalluSquatting: Researchers Turn AI Hallucinations Into a Malware Delivery Channel

Researchers have demonstrated a technique called "adversarial hallucination squatting" — or HalluSquatting — that exploits the tendency of AI assistants to confidently recommend non-existent package names. The attack works by registering the fake package names that popular AI coding assistants frequently hallucinate, then stuffing them with malware. When a developer asks an AI for help and it suggests a package that doesn't exist, an attacker who pre-registered that name gets code execution on the developer's machine. It's a new flavour of typosquatting that requires no typo — just a predictably wrong AI. The technique has been demonstrated against multiple popular AI assistants and achieves remote code execution. This is a genuine supply chain risk for any team using AI-assisted coding without package verification discipline.

SecurityWeek

SK Hynix Raises $26.5 Billion in the Biggest Foreign IPO in US History

South Korean memory chipmaker SK Hynix opened on Wall Street at $170 per share on Friday, raising $26.5 billion and shattering Alibaba's record for the largest US debut by a foreign company. The AI boom's insatiable appetite for high-bandwidth memory — the specialised RAM that makes Nvidia's GPUs sing — has turned SK Hynix into one of the most strategically critical companies in the global chip stack. US officials used the debut to press SK Hynix and Samsung to build domestic American fabs. The trillion-dollar valuation and Wall Street reception underline just how much the AI infrastructure story is really a memory story. For Australian superannuation funds with US equity exposure, SK Hynix is now a name that matters.

TechCrunch

GigaWiper: The Swiss Army Knife of Destructive Malware

Microsoft has detailed GigaWiper, a modular Windows backdoor that bundles at least three distinct malware families into a single package: a standalone wiper, ransomware-grade encryption, and a multi-pass data destruction command. The modular design means attackers can switch objectives mid-intrusion — start with reconnaissance, pivot to encryption for ransom leverage, then trigger the wiper if negotiations fail or they simply want to cover tracks. SecurityWeek and The Register both covered the disclosure, noting it represents a meaningful step up in destructive capability from typical ransomware tooling. The combination approach is particularly nasty for incident responders, who may face simultaneous recovery and negotiation decisions with no clean path through either.

SecurityWeek

Free Android VPN Apps Are Leaking the Traffic They Promise to Protect

A study testing 281 of the most popular free VPN apps on the Google Play Store found widespread, basic failures: 29 apps leak user traffic outside the VPN tunnel entirely, and many transmit unencrypted data or embed tracking SDKs. The apps flagged with at least one problem have been installed more than 2.4 billion times combined. The failures aren't sophisticated — they're sloppy implementations of fundamentals. This matters particularly for users in high-risk environments who believe a free VPN is protecting their traffic. In Australia, where free VPN use spiked after various geo-blocking discussions, this finding is a direct consumer warning: the free VPN you installed is quite possibly doing the opposite of what you think.

The Hacker News

EU Threatens Meta With Fines Over Autoplay and Infinite Scroll

The European Commission has formally told Meta that autoplay video, infinite scroll, push notifications, and its personalised recommendation algorithms breach the Digital Services Act — and that fines are coming if the company doesn't make changes. The DSA requires platforms to assess and mitigate risks of "addictive design" on minors, and the Commission has decided Meta has failed that test. Meta will almost certainly challenge the ruling, but the regulatory pressure is real. Australia is watching closely: the Online Safety Act's Basic Online Safety Expectations and the ongoing reform of age assurance rules put Australian regulators on a parallel track, and EU enforcement precedents tend to inform what ACMA considers politically achievable domestically.

TechCrunch

CBA Cuts 176 Technology and Engineering Roles

Commonwealth Bank of Australia is cutting 176 technology and engineering positions, the bank confirmed to iTnews. CBA pushed back on suggestions the roles are being offshored, but offered little detail on where the work goes instead. The cuts land at an interesting moment: CBA has been one of the more aggressive Australian financial institutions in publicising its AI investments, and headcount reductions in engineering during an AI buildout tend to tell a story about where productivity gains are actually landing. Whether this is AI-driven efficiency, cost discipline ahead of a softer economic period, or both is worth watching. It's also a significant development for Australia's tech employment market, where financial services engineering roles carry real weight.

iTnews

Ransomware Negotiator Who Betrayed His Own Clients Gets 70 Months

Angelo Martino, a former ransomware negotiator at DigitalMint, has been sentenced to 70 months in federal prison after prosecutors established he was secretly feeding confidential victim information to the BlackCat/AlphV gang he was supposed to be negotiating against. Martino and two co-conspirators helped extort a combined $75.3 million from five US-based victims. He is the third security professional sentenced in relation to ransomware conspiracy charges in recent months, a pattern that signals US prosecutors are actively pursuing insider collusion as a category, not just treating it as an oddity. The case is a sharp reminder that incident response — including negotiation — creates privileged access that can be catastrophically abused.

CyberScoop

Patreon Partners With Cloudflare to Block AI Crawlers From Creator Content

Patreon has partnered with Cloudflare to block AI training crawlers from scraping creator content, with CEO Jack Conte drawing a hard line: "Creators deserve credit, compensation, and consent. If that's not on the table, the crawlers can stay the fuck off Patreon." The move follows similar actions from other platforms and reflects growing creator-economy pressure on AI labs to negotiate data licensing rather than scrape freely. Cloudflare's AI crawler blocking tools have been increasingly adopted by publishers and platforms globally, including several Australian media organisations that have quietly deployed them over the past year. Whether this constitutes meaningful protection or a speed bump that determined crawlers route around remains an open question.

404 Media

Sources consulted