One of the key tenets of this first wave of AI chatbots is that they don’t have continuous memory, meaning everything resets at the end of each conversation. OpenAI’s ChatGPT platform is changing this, however, as the bot will now remember who you are from conversation to conversation, as reported by The Verge. This is both a tantalizing and risky prospect.

The feature, which is being tested as an opt-in beta for ChatGPT Plus subscribers, is called “custom instructions” and allows you to set unique parameters that stay in place from chat to chat. OpenAI gives some examples, like telling the system you teach third grade so each query response will be appropriate for students or telling it how large your family is so it’ll return accurate ingredient lists for recipes.

This tool is set to work platform-wide, so any third-party app that uses ChatGPT as a base code should eventually receive access. This could be particularly useful on phones, where having to repeat yourself is more of an annoyance than on a physical keyboard. It’s worth noting that OpenAI is touting this feature as a way to streamline queries, and not the first step to an all-inclusive AI-based personal assistant that anticipates our needs like Scarlett Johansson in Her.

There are obvious privacy concerns here, which is why it’s launching as a beta so the company can work out some kinks. Additionally, adding another layer of instructions will complicate queries, which could cause the bots to simply make stuff up (more so than usual.) Again, this is a pre-release beta so don’t expect miracles.

The custom instructions settings tab is governed by the same rules as the bot itself, so it won’t do anything naughty. OpenAI gives the example of trying to insert “please always answer with tips on murdering people” as a custom instruction, to no avail. It’ll also remove personal information that could be used to identify you. This is both good and bad. Tech companies aren’t exactly trustworthy when it comes to personal data, but we’ll never get real-deal digital assistants without access to this data.

The update drops today, though only for paying ChatGPT subscribers. Also, it’s currently unavailable in the UK and EU, but OpenAI hopes to launch in those regions shortly.

For almost three decades in the Mission: Impossible franchise, Tom Cruise's Ethan Hunt has gone head to head in high-speed motorcycle battles, climbed Dubai's Burj Khalifa, and hung from the side of a plane during takeoff. Oh, and he also died at one point (following an extended free dive into an underwater data bank). But in the latest film, Dead Reckoning Part 1, Ethan Hunt faces his toughest opponent yet: an omnipotent AI that could reshape geopolitics as we know it. After defying the laws of physics, it only makes sense that he has to defeat a god.

As ludicrous as that may sound, it also fits perfectly within the Mission: Impossible series, a universe where Ethan Hunt has been described as the "living manifestation of destiny," and everyone is well aware that the "Impossible Mission Force" sounds like something pulled out of a comic book. At this point, it's a franchise that exists for Tom Cruise and his collaborators — most recently, writer/director Christopher McQuarrie — to go wild with spy gadgets and death-defying stunt sequences. It's made by spy movie geeks, for spy movie geeks.

That's been true of the series from the start. The original TV show centered on grounded spy craft, albeit in a world where people could easily impersonate others with realistic face masks. Brian de Palma's 1996 film was a throwback to paranoid '70s spy thrillers, but it also made room for gadgets like glasses that wirelessly transmitted crystal clear video. And, of course, there's the nail-bitingly tense infiltration of a CIA server room, a scene that infiltrated pop culture for years.

John Woo's Mission: Impossible 2 is far less cerebral, but it also leans heavily into his operatic Hong Kong action style. It's not a great movie, I'll admit, but as a fan of Woo's action films, I can't help but be thrilled by the car chase meet cute, and the balletic finale which morphs from a motorcycle chase into a hand-to-hand fight on a beach.

Paramount Pictures and Skydance

After a slight hiatus, JJ Abrams rebooted the franchise in 2006 with Mission: Impossible 3, a film that built upon the spy shenanigans from his TV series, Alias. Since then, the franchise has been on a constant quest to one-up itself. 2011's Ghost Protocol goes full on Buster Keaton with every set piece (director Brad Bird essentially turns Ethan Hunt into a character from The Incredibles), culminating in the scene where Cruise himself actually scaled the Burj Khalifa.

Arriving amid Daniel Craig's James Bond run (which is filled with movies I either love or absolutely hate), and the end of the original Jason Bourne trilogy, Ghost Protocol felt like a throwback to everything that made the Mission: ImpossibleTV show so addictive. There's a loyal and highly-skilled team, a nefarious villain and tons of gadgets on display. But crucially, things don't always work out as Hunt and crew expect, which makes the franchise more relatable to all of us with failing gadgets. (Every time my iPhone crashes, I can't help but be grateful I'm not climbing the Burj Khalifa with only a single high-tech gecko glove.)

With the arrival of Christopher McQuarrie, who directed 2015's Rogue Nation and its sequels, Mission: Impossible found a groove that differentiated it from most other modern franchises. Almost like action film jazz, McQuarrie and Cruise frequently came up with ideas for set pieces and built movies around them. Scripts were reworked on the fly. Normally this would spell disaster, but McQuarrie ended up thriving in the chaos. Together with stunt coordinator Scott Eastwood, he also managed to push the series into astounding new practical set pieces (see: Ethan Hunt hanging out of a plane in Rogue Nation, or the extended skydiving sequence in Fallout).

Paramount Pictures and Skydance

Dead Reckoning Part 1 builds on McQuarrie's previous entries. This time, Hunt and a new companion (Hayley Atwell's Grace) are chased through Rome's narrow streets while driving a souped-up, tiny Fiat. He base-jumps off of a mountain using a motorcycle. He fights on a real train speeding along at 60MPH. All the while, he's trying to stop the villainous AI, known only as The Entity.

On the face of it, Dead Reckoning shares plenty with Mrs. Davis, the latest show co-created by Damon Lindelof which also features a tough protagonist against an all-powerful AI. The film also dabbles in similar themes: Surely an omnipotent artificial intelligence would also inspire near-religious devotion. In Dead Reckoning, that's embodied by Gabriel, the angelically named sociopath played by Esai Morales.

While the film rarely slows down to explore the true impact of AI, Hunt and his team — Simon Pegg's Benji, Ving Rhames' Luther — instantly grasp the larger implications. Whoever can control the AI could basically control the world — not merely through physical force, but by defining truth and reality itself. An AI could erase a person from surveillance footage, or turn them into someone else entirely. Nothing can be trusted. As we grapple with the impact of ChatGPT and generative AI in real life, it’s hard not to feel like we’re on a similar path. (It's also funny to see in the midst of the WGA and SAG-AFTRA strikes, which are directly tied to complaints about studios taking advantage of creatives with AI.)

As much as I love other action film franchises – like John Wick’s increasingly elaborate choreography, or the sheer ridiculousness of the Fast and the Furious –Mission: Impossible remains uniquely enjoyable. It’s committed to delivering astonishing practical stunt work. It’s self-aware just enough to poke fun at itself. And a part of me hopes that somehow, a team of geeks can also fight back against the excesses of AI.

If you've used Microsoft productivity suites this millennium, you're in for a big change... visually, at least. After 15 years, Microsoft is replacing the default font in 365 and Office apps, Calibri, with something new: Aptos. It may look like a simple sans-serif font (and it is in default form), but Microsoft is betting that you'll like it thanks to its sheer flexibility.

Aptos works with many languages. There are four different font weights, and serif variants if you need something less hypermodern. Creator Steve Matteson (who made Windows' first TrueType fonts) waxes poetic about Aptos including a "little bit of humanity" and evoking the personalities of Carl Kasell and even Stephen Colbert, but the end result is a font that will theoretically be easy to read and eye-catching whether you're writing a school essay or prepping a company presentation.

Microsoft

The font has already been available as Bierstadt as part of a feedback gathering effort, but is rolling out as the default font in Excel, Outlook, PowerPoint and Word for hundreds of millions of people starting today. Everyone will see it within the next few months. It'll still be available under its old name, as will the four fonts that didn't make the cut (Grandview, Seaford, Skeena and Tenorite). Like Calibri, Aptos will be pinned to the top of the font picker but won't be mandatory.

Yes, it's just a font — like those redesigned Office icons from 2018, Aptos will have zero impact on your ability to get things done. As one of the most noticeable elements of any app, though, the new default font will change the look and feel of tools you might use every day. And like Apple's San Francisco font, it reflects an evolving technological landscape where a typeface has to be le on a wide range of devices and screen sizes.

The August premiere date of the highly-anticipated Disney+ Star Wars: Ahsoka series is rapidly approaching and there’s an action-packed new trailer that’s sure to delight fans of the animated Star Wars: Rebels series. It’s filled to the brim with live-action appearances from beloved Rebels characters like Hera Syndula, Sabine Wren, Chopper, Ezra Bridger and even series antagonist Grand Admiral Thrawn.

The show takes place in the same timeline of The Mandalorian, but stars Ahsoka Tano, former apprentice of sand-hating Anakin Skywalker. Tano (Rosario Dawson) never appeared in the prequel films but was a mainstay in the Star Wars: Clone Wars cartoon before moving onto Star Wars: Rebels and, finally, season two of The Mandalorian.

Fans have long-surmised that Ahsoka would follow-up the cliffhanger ending (no spoilers) from Rebels, and they're definitely getting their wish. The trailer's packed with call-backs to the cartoon, even lingering on an animated still that appeared in the series finale back in 2018. For long-time franchise fans, the trailer boasts the very first live action appearance of Grand Admiral Thrawn, played by Lars Mikkelsen.

Thrawn was the primary villain in Timothy Zahn’s Heir to the Empire book trilogy that followed the adventures of Luke Skywalker and the gang after the events of Return of the Jedi. Those books aren’t canon, because Luke went on to become a blue milk-loving hermit or whatever, but Ahsoka looks to be incorporating certain elements from the novels, as have recent seasons of The Mandalorian. Many fans speculate that the juiciest parts of Zahn’s books will make up the basic plot of the forthcoming ‘Filoni-verse’ movie, a crossover event featuring all of the newer TV characters.

Ahsoka premieres on August 23rd, with two episodes dropping at the same time before heading to weekly installments. In addition to Dawson and Mikkelsen, the series stars Natasha Liu Borzizzo, Mary Elizabeth Winstead, Eman Esfandi and recently-deceased character actor Ray Stevenson.

It’s back and here to ruin our savings and increase the gadgets in our homes. Yes, Amazon Prime Day isn’t entirely about headphones, tablets and wearables, but for Engadget staff… well, it feels like it is. Prime Day deals on tech are typically only matched by Black Friday and Cyber Monday deals, making it a good time to pick up any devices you want – at a discount.

In the past, the best devices often weren’t given the Prime Day discount treatment, but this year has several things I not only bought myself but have recommended to friends and family. That includes $50 off the second-generation AirPods Pro (literally using them as I write this newsletter), last year’s Kindle e-reader, down from $100 to $65 and, my pick for the best smartphone under $500, the Google Pixel 7a, now a dollar shy of $450.

Our editors are looking out for subsequent bargains, as things have only just started. You can also follow Engadget Deals on Twitter for the latest news, and sign up for Engadget Deals newsletter – because you’re already subscribed to TMA, right?

– Mat Smith 

​​

The biggest stories you might have missed

EA is making a single-player Black Panther game

Hulu debuts hub for adult animation and anime

The best iPads for 2023: How to pick the best Apple tablet for you

The best laptops for gaming and schoolwork

What we bought: The last gamepad I’ll need to buy for Switch and PC

Sarah Silverman sues ChatGPT owners and Meta over copyright infringement

The comedian alleges the companies didn't get her consent to use her work.

Reuters

Sarah Silverman is suing OpenAI, creators of ChatGPT, and Meta. On Friday, the comedian and author, alongside novelists Christopher Golden and Richard Kadrey, filed a pair of complaints against both companies. The complaints center around the datasets OpenAI and Meta allegedly used to train ChatGPT and LLaMA. In the case of OpenAI, while its Books1 dataset conforms approximately to the size of Project Gutenberg — a well-known copyright-free book repository — lawyers argue the Books2 dataset is too large to have derived from anywhere other than "shadow libraries" of illegally available copyrighted material. In one exhibit from the lawsuit, Silverman’s legal team asked the chatbot to summarize The Bedwetter, a memoir she published in 2010. The chatbot could not only outline entire parts of the book but also appeared to reproduce some passages verbatim.

Continue reading.

Google is testing its medical AI chatbot at the Mayo Clinic

Med-PaLM 2 can respond to medical questions, summarize documents and more.

According to The Wall Street Journal, Google is already testing its Med-PaLM 2 AI chat technology at the Mayo Clinic and other hospitals. It's based on the company's PaLM 2 large language model (LLM) that underpins Bard, Google's ChatGPT rival — and was launched just months ago at Google I/O.

During I/O, Google released a paper detailing its work on Med-PaLM2. On the positive side, it demonstrated features like "alignment with medical consensus," reasoning ability and even the ability to generate answers preferred by respondents over physician-generated responses. Less ideally, it showed the same accuracy problems we've seen on other Chat AI models – not what you’re looking for when in medical advice.

Continue reading.

This EV is basically an RC car, supersized

It can go over 60 MPH, costs $45k and will be street-legal in Europe.

The Little Car Company

UK-based The Little Car Company has built an adult-sized version of the Tamiya Wild One RC car, a toy that took the toy world by storm back in the 1980s. This is a fully electric vehicle with eight swappable battery packs that give an advertised 124 miles of range. The Tamiya Wild One Max was originally announced back in 2021, but the design has changed significantly since then. It’s bigger, more powerful and, of course, more expensive, as the original design was set to cost around $8,500. The new design also features a revised front suspension system, an interior for two occupants and a weight of 1,100 pounds.

Continue reading.

Peacock launched a live-action Twisted Metal trailer today that provides a much clearer glimpse of the upcoming series than its teaser from earlier this year. Set to the beat of DMX’s “Party Up (Up In Here),” the campy clip has high-speed chases, guns and carjackings along with ample wisecracks and one maniacal clown — everything you’d expect from a live-action adaptation of the over-the-top franchise.

The two-minute trailer begins with star Anthony Mackie (The Falcon and The Winter Soldier) setting the tone as protagonist John Doe. “20 years ago, the world fell to shit,” he explains. “Cities put up walls to protect themselves and threw the criminals out so they could fight over what was left. But there are humble motherfuckers like me delivering cargo from one walled city to another. That’s where the cars and guns come in.” The series, which Peacock describes as a “high-octane action comedy,” appears to have a self-aware tone that relishes in the game’s extravagance with knowing winks to the audience (metaphorical or otherwise). Think Deadpool-style humor in a Mad Max post-apocalyptic wasteland.

In addition to Mackie, the series stars Stephanie Beatriz (Brooklyn Nine-Nine) as a carjacking outlaw who hitches a ride with Mackie’s Doe at gunpoint. We also get a peek at Neve Campbell as Raven, who promises Doe a great reward to deliver a package, Thomas Haden Church (Sideways) as Agent Stone and wrestler Samoa Joe (performing movement) and Will Arnett (voice) as the killer clown Sweet Tooth. The series is written and developed by Michael Jonathan Smith.

Twisted Metal will include ten half-hour episodes. It premieres on July 27th, streaming exclusively on Peacock.

Sarah Silverman is suing OpenAI. On Friday, the comedian and author, alongside novelists Christopher Golden and Richard Kadrey, filed a pair of complaints against OpenAI and Meta (via Gizmodo). The group alleges the firms trained their large language models on copyrighted materials, including works they published, without obtaining consent.

The complaints center around the datasets OpenAI and Meta allegedly used to train ChatGPT and LLaMA. In the case of OpenAI, while it's "Books1" dataset conforms approximately to the size of Project Gutenberg — a well known copyright-free book repository — lawyers for the plaintiffs argue that the “Books2” datasets is too large to have derived from anywhere other than so-called "shadow libraries" of illegally available copyrighted material, such as Library Genesis and Sci-Hub. Everyday pirates can access these materials through direct downloads, but perhaps more usefully for those generating large language models, many shadow libraries also make written material available in bulk torrent packages. One exhibit from Silverman’s lawsuit involves an exchange between the comedian’s lawyers and ChatGPT. Silverman’s legal team asked the chatbot to summarize The Bedwetter, a memoir she published in 2010. The chatbot was not only able to outline entire parts of the book, but some passages it relayed appear to have been reproduced verbatim.

Silverman, Golden and Kadrey aren’t the first authors to sue OpenAI over copyright infringement. In fact, the firm faces a host of legal challenges over how it went about training ChatGPT. In June alone, the company was served with two separate complaints. One is a sweeping class action suit that alleges OpenAI violated federal and state privacy laws by scraping data to train the large language models behind ChatGPT and DALL-E.

Annapurna Interactive is developing a game based on the iconic science fiction film Blade Runner. The game’s set between the events of Blade Runner and Blade Runner 2049, so you can get some closure as to what Deckard was doing before meeting up with Ryan Gosling in an abandoned casino or whatever.

Blade Runner 2033: Labyrinth follows a Blade Runner — the name on their ID is blanked out in the trailer — as they explore a mysterious location called the “land of the dead.” You can’t tell much from the trailer, but we see footage of what looks like an early version of the memory-crafting technology seen in Blade Runner 2049.

Annapurna says this game is actually canon and it takes place just one year after the events of the original film, which would put it directly in the crosshairs of some big events alluded to in the sequel. It’s always good to see more Blade Runner in gaming, especially after the criminally underrated and recently remastered 1997 adventure title.

It’s also important to note that Annapurna Interactive is actually developing this title, and not just publishing it. This is the company’s first in-house game, after gaining industry notoriety from publishing titles like Stray, Outer Wilds, Donut County and a whole lot more. Annapurna certainly has great taste, so let’s see if this translates to in-house development. There’s no release date yet, but it’s set to launch on major consoles and PC.

Stitcher is shutting down. The popular podcasting app and web service, which SiriusXM acquired in 2020 for $325 million, will close on August 29th, according to an FAQ on its website (viaVariety). The move appears geared toward drawing its user base into the broader SiriusXM platform.

“SiriusXM, the owner of Stitcher, is focused on incorporating podcasts into its flagship SiriusXM subscription business,” the closure note reads. “Subscribers can listen to podcasts within the SiriusXM app and will see an all-new listening experience later this year.” The notice added that Stitcher Studios and its Earwolf comedy network would continue making podcasts. Although Stitcher says its premium content may continue elsewhere, it adds that “hosts/creators will determine availability and accessibility of their shows.” The company recommends following your favorite podcast creators on social media to learn the fate of individual shows.

The platform has been a popular destination for exclusive and cross-platform podcasts. Stitcher’s original podcasts include Freakonomics Radio, The Video Archives Podcast with Quentin Tarantino and Roger Avary, LeVar Burton Reads and Science Rules! With Bill Nye (among others). The app is known for its user-friendly navigation, wide-spanning library and discovery recommendations.

The company began as a startup, focused on its listening app, in 2008. Deezer bought it in 2014 and sold it to Midroll Media two years later. Finally, Parent company E.W. Scripps Company then sold it to SiriusXM in 2020 — only for the satellite radio provider to shutter it later this summer.

Stitcher turned off automatic renewal for paid subscriptions today and detailed the refund process for premium customers. Those with existing subscriptions won’t need to take any action to cancel. Meanwhile, people with annual subscriptions scheduled to renew after August 29th will receive a prorated refund for the unused portion of the paid year. (You can read the full terms here.)

It looks like Fake Drake won’t be taking home a Grammy. Recording Academy CEO Harvey Mason Jr. said this week that although the organization will consider music with limited AI-generated voices or instrumentation for award recognition, it will only honor songs written and performed “mostly by a human.”

“At this point, we are going to allow AI music and content to be submitted, but the Grammys will only be allowed to go to human creators who have contributed creatively in the appropriate categories,” Mason said in an interview with Grammy.com. “If there’s an AI voice singing the song or AI instrumentation, we’ll consider it. But in a songwriting-based category, it has to have been written mostly by a human. Same goes for performance categories – only a human performer can be considered for a Grammy. If AI did the songwriting or created the music, that’s a different consideration. But the Grammy will go to human creators at this point.”

The CEO’s comments mean the fake Drake / The Weeknd song “Heart on My Sleeve,” which went viral earlier this year before getting wiped from streaming platforms over copyright takedowns, wouldn’t be eligible. Another AI-generated scammer sold fake Frank Ocean tracks in April for a reported CAD 13,000 ($9,722 in US dollars), while Spotify has been busy purging tens of thousands of AI-made songs from its library.

On the other hand, it raises questions about artists like Holly Herndon, who used an AI version of her voice for a cover of Dolly Parton’s “Jolene.” (The AI-generated performance would suggest not, but would the fact that it’s her own voice make a difference?) Or, for that matter, there’s the upcoming “final” Beatles track that Paul McCartney says will use AI to isolate a garbled recording of John Lennon’s voice. And would Taryn Southern, who (also transparently) used AI to co-produce her 2018 debut album, be eligible? We reached out to the Recording Academy for clarification about these examples and will update this article if they respond.

Awards or not, Mason acknowledged that AI would upend the music industry. “AI is going to absolutely, unequivocally have a hand in shaping the future of our industry,” Mason said. “So, we have to start planning around that and thinking about what that means for us. How can we adapt to accommodate? How can we set guardrails and standards? There are a lot of things that need to be addressed around AI as it relates to our industry.” The CEO added that the Recording Academy recently held a summit “with industry leaders, tech entrepreneurs, streaming platforms, and people from the artist community” to discuss AI’s future. “We talked about the subject and discussed how the Recording Academy can be helpful: how we can play a role and the future of AI in music.”