Alibaba told staff to stop using Anthropic's Claude Code from 10 July, citing fears the tool can flag China-linked users and quietly collect data like timezones and proxy settings. Employees are being pushed towards Qoder, Alibaba's homegrown alternative. The ban lands days after Anthropic accused Alibaba of ripping off Claude's capabilities to train its own models — so this isn't really about one coding tool. It's about two AI giants who no longer trust each other's software, and a signal that the AI race is splitting into rival, mutually suspicious camps.

Christine Lagarde admitted she wouldn't rule out leaving the European Central Bank before her term ends in October 2027, if it meant she could play a role in France's next presidential race. She insists she has no plans to run — but "no plans" has never stopped speculation before. The comment matters because central bank independence relies on leaders looking uninterested in politics, not adjacent to it. If Europe's top rate-setter is even entertaining an early exit, markets will start pricing in who comes next, and whether Frankfurt's next resident will be quite so steady.

Mark Zuckerberg told staff at an internal town hall that Meta's AI agents are progressing more slowly than expected, despite billions poured into infrastructure and a major team reshuffle. He's still committed to the long-term "superintelligence" goal, but admitted meaningful results are three to six months out, not now. For a company that's bet its next act on AI catching up to its own hype, an admission like that from the top is notable. It's a reminder that even the deepest pockets in tech can't just buy their way past the hard parts of building genuinely useful AI.

UK digital bank Starling is cutting 130 roles, about 3% of its workforce, as part of a push to simplify operations and lean harder on artificial intelligence. The bank will keep hiring AI and tech engineers even as it trims elsewhere, following a dip in annual revenue and profit. It's the same pattern showing up across fintech: growth alone no longer buys goodwill, and efficiency is the new pitch to investors. For staff, it's a blunt reminder that "AI-powered simplification" is corporate-speak that often starts with headcount.

Goldman Sachs advised on 44% of all M&A deal value across Europe, the Middle East and Africa in the first half of 2026, as regional dealmaking hit a 19-year high of $676 billion. It worked on 15 of the region's 20 biggest deals, pulling further ahead of JPMorgan and other rivals. The boom reflects companies chasing long-term strategic bets despite a choppy market backdrop. For smaller advisory firms, the message is blunt: when the biggest deals are on the table, one bank is increasingly the only name in the room.

Ben Francis is reportedly in talks to buy back part of the 21% stake in Gymshark he sold to private equity firm General Atlantic back in 2020. The brand's revenue is still climbing, but profits are under pressure as competition in activewear intensifies. Francis clawing back control fits a wider mood among founders who sold early and now want more say over where their company goes next. It's a reminder that private equity deals aren't always a one-way street — sometimes the person who built the thing wants the wheel back.

Global hiring is changing fast. From AI's impact on HR to evolving compliance requirements and international expansion strategies, the rules are constantly shifting.

Oyster's events bring together HR leaders, founders, operators, and global employment experts to discuss what's working now—and what's coming next.

Whether you're hiring internationally today or planning for tomorrow, you'll walk away with practical insights you can actually use.

Explore Hiring Events

A year ago, Donald Trump announced the US had struck a trade framework with Vietnam: a 20% tariff on most Vietnamese exports and a steep 40% levy on goods transshipped through Vietnam from other countries, arriving days before a tariff deadline was due to bite. Nike and other Vietnam-reliant retailers rallied on the news. A year on, with the framework still being finalised and transshipment rules still contested, it's a reminder that "done deals" in trade policy rarely stay done.

What it means: It's when one AI company trains its own model by feeding it outputs from a rival's model — essentially learning the answers without doing the underlying research. It's cheaper and faster than building from scratch, but it's also the accusation at the heart of Anthropic and Alibaba's current stand-off, and one reason Alibaba's staff are being steered away from Claude Code altogether.