President Donald Trump signed three proclamations on Monday imposing 50% tariffs on a range of Canadian goods — wine, dairy, cement, furniture and even ice-hockey gear — though energy, potash and critical minerals were spared. To do it he reached for Section 338 of the Tariff Act of 1930, an obscure Depression-era statute that lets a president punish nations deemed to discriminate against American commerce. The duties, covering some $20bn of trade, take effect in thirty days.

The White House argues Canada has long tilted the playing field, pointing to Canadian levies on American cars, milk and drink. Ottawa sees it differently: Prime Minister Mark Carney called the move a breach of the USMCA free-trade pact, yet pointedly declined to retaliate, offering instead to “engage intensively” in talks. Provincial premiers were fierier, demanding a response “dollar for dollar”. Economists warn that tariffs ultimately fall on consumers, stoking inflation on both sides of the border.

Harvard engineers have turned an ordinary silicon chip into a miniature DNA factory. Writing in Nature Electronics, the team led by Donhee Ham describes a device that synthesises 64 different strands of DNA at once, using only electric currents and water-based enzymes. Tiny ring-shaped electrodes fire protons at precise spots on the chip, nudging each strand to grow letter by letter — a far cleaner method than the solvent-heavy chemistry that dominates DNA manufacturing today.

The promise is tantalising: portable machines that print bespoke genetic material for medicine or vaccines, and perhaps one day vast archives of data stored in DNA itself. Yet caution is warranted. The process still struggles to scale, hampered by stray molecules that blur the chemistry, and writing genetic code cheaply raises thorny questions about biosecurity. Building the very alphabet of life, it turns out, is becoming an engineering problem.

This summer, the question of who owns the words and images that train artificial intelligence is coming to a head in Britain. Getty Images, one of the world's largest photo agencies, is appealing a landmark High Court ruling from last November that largely cleared the AI firm Stability AI of copyright infringement. The judge found that training a model on protected pictures did not, in itself, breach UK copyright, because the finished model stores no copies of the originals.

Writers, photographers and musicians say this leaves them exposed: a lifetime of work can be scraped to build lucrative tools without their consent or a penny in payment. AI companies reply that learning from publicly available material is no different from a human studying it, and warn that heavy restrictions would simply push innovation abroad. Ministers are now testing a licensing "exchange" to let creators be paid — an attempt to reconcile two industries that each claim to be the future.