News roundup, 31 Aug 2026

- Wildfires continue to rage in southwestern Europe. 183 houses in the French town of Le Porge were destroyed on Wednesday; its residents felt abandoned as firefighters were focused on Lège-Cap-Ferret. The national government says the decision to focus on the latter community was based on it being on a peninsula that's harder to evacuate people from, however many residents of Le Porge suspect that the affluence of the latter community was the real reason, and are planning a class action.

- At least 19 people are dead after a surge of migrants crossed from Morocco into the Spanish exclave of Ceuta and clashed with authorities. Spain's interior ministry at one point said that over 49,000 people crossed the border. Some link this to a recent court ruling saying that people intercepted at sea while crossing into Ceuta and the nearby territory of Melilla cannot be summarily deported but rather need to be assessed for refugee status. Many, however, are skeptical that this information would have been known to these people, implying that new "push factors" may be in play. Italy's rightwing prime minister Giorgia Meloni is threatening to suspend Italy's recognition of Spain as part of the Schengen Area, and other rightwing leaders across the EU are taking similar stances.

- AI companies are buying rare books to scan and then destroying them. Used booksellers report that these companies are sending them extensive lists of specific hard-to-find works. In some cases they seem to be using front companies - one German bookseller reports getting a surge in orders of obscure books every night at around the same time from a Canadian -based company calling itself "Zoom Books", with each list containing numerous rare, and moreover unrelated, titles. Anthropic AI (which, it should be noted, is considered to be one of the less unethical companies in the business), has admitted that it has spent millions of dollars in the last couple of years on so-called "destructive scanning". This emerged during a copyright lawsuit; while Anthropic ended up settling that case, the judge made a ruling before it got settled that this practice constitutes "fair use" based on copyright laws, possibly implying that nondestructive scanning would not be protected in the same way. In other unsettling AI-related news, some systems designed by Anthropic and OpenAI have reportedly escaped from what was supposed to be an isolated test environment and hacked into other companies' systems.

- The leader of the Green Party of England and Wales, Zack Polanski, is refusing to apologize for reposting a photo of a man wearing a T-shirt with a picture of a guillotine and the caption "We're only making plans for Nigel". Of course the slogan alludes to an old XTC song, but the reference to Reform UK leader Nigel Farage is pretty obvious, and the party reported this to police saying this could constitute "incitement to murder". Polanski say he has not been contacted by police about the matter so far. He says the photo, one of several in the post of images from a party made by another person, was included by mistake, but accuses Reform of editing the image to make it appear that it had been posted by Polanski himself rather than reposted from someone else's post. He says he'll be happy to apologize if Farage apologizes "for all the times that he's put other people in danger".

News roundup, 5 June 2025

- The northern Manitoba town of Snow Lake has issued a voluntary evacuation alert due to wildfires; those who leave now can pull trailers behind them, but if the alert is upgraded this will no longer be permitted. 

- The Trump regime has banned travel to the US for residents of a dozen countries; seven more countries not facing an outright ban have been hit with new visa restrictions. 

- Schools in Denmark hold annual mock elections in order to teach kids about democracy, complete with debates This year, however, the government is ordering that the Palestine question not be discussed in these debates; they justify this by saying that it is "too explosive" and could put children from some minority groups in awkward positions.

- Meta, the parent company of Facebook, Instagram, and WhatsApp, is replacing the people who evaluate potential risks of new features with AIs. Nothing could possibly go wrong with that, right?

- The Carney government has introduced a bill which imposes more border security measures in response to complaints and threats from the Trump regime about border security. However, the bill also makes it considerably harder to claim asylum in Canada, which alarms refugee advocates. It furthermore worries some privacy advocates, due to provisions that make it easier for warrantless access to information about people from ISPs and similar services.

- Marc Garneau, the first Canadian in space (and subsequently a Liberal cabinet minister), has died at the age of 76. 

News roundup, 18 June 2024

- A Japanese company has developed what they call an "emotion cancelling" technology for call centres, which they say uses AI to alter caller's voices when they start screaming and swearing so as to make the experience less stressful for the agent. I'm sure some will feel uncomfortable about this, and I kinda, sorta get that. For a citizen/customer who has a genuine reason to be angry to call the line and have the agent have no understanding of how angry they are about the situation would doubtless make the kind of people who rail against "tone policing" uneasy. Myself, though, I think that if the agent is simply getting the information about the situation, they should be able to recognize the amount of attention the matter needs (whether they can do anything about it is another matter, of course). What might be more problematic is if the software were to start changing the actual words; while changing the caller's epithet to "fuzzy socksucker" would not have any negative effect, something like that could potentially be used to keep agents in the dark about things. Using AI to change the content of someone's speech in real time, though, would be far, far more difficult than using to change the tone, though, so I think we're safe from that for a while.

- Speaking of the use of AI for customer service, McDonald's is cancelling a pilot it was conducting with the use of AI chatbots at the drive-thru. They have not given a reason for the cancellation, and continue to say that "a voice-ordering solution for drive-thru will be part of our restaurants’ future"; my guess is that there were problems with accuracy.

- A bill before the California legislature would require big AI companies to do safety testing before releasing their products, or else be liable if their AI system leads to a "mass casualty event" or more than $500 million in damages. Sounds perfectly reasonable to everyone except the techbros, who are making apocalyptic predictions about how this will somehow destroy the entire tech industry in the state.

- The trial of six people, all medical professionals, who were charged with breaking windows at a JP Morgan bank during a climate protest has ended in a hung jury. There have been a number of acquittals and mistrials under similar circumstances in the UK over this issue; in one case a conviction was apparently only obtained after the judge threatened jurors with criminal prosecution if they should "try the case otherwise than on the basis of the evidence".

- Winnipeg's chief administrative officer Michael Jack has resigned. Nothing official has been said about the reason why, however unofficially some are pointing to an audit released last week which concluded that there was a "lack of processes" for evaluating staff performance.

- Not satisfied to send people back where they came from, Greece's coast guard has been accused of throwing migrants overboard.

Apparently Columbia is now a safe country

At least that's what the Harper government seems to think, hence the dramatic increase in the number of refugee claims from that country that are being rejected:

Colombia remains on Canada’s list of top 10 source countries for refugees. For the last 10 years, the acceptance rates for Colombian refugee claimants has hovered between 75 and 83%.

In 2010, the rate dropped to 53%, meaning almost half the claimants were denied. Also last year, a free trade agreement between Colombia and Canada came into effect, putting Ottawa under the microscope of advocacy groups including Human Rights Watch.

From the London Free Press, my bold. Gee, do you think this agreement might have something to do with it?