Microsoft is updating Outlook for Mac to include one of the best new features Apple has added to Mail in recent years. The update also brings Outlook for Mac up to speed with the Windows and web version when it comes to this capability.
more…
Even the most permissive corporate AI models have sensitive topics that their creators would prefer they not discuss (e.g., weapons of mass destruction, illegal activities, or, uh, Chinese political history). Over the years, enterprising AI users have resorted to everything from weird text strings to ASCII art to stories about dead grandmas in order to jailbreak those models into giving the "forbidden" results.
Today, Claude model maker Anthropic has released a new system of Constitutional Classifiers that it says can "filter the overwhelming majority" of those kinds of jailbreaks. And now that the system has held up to over 3,000 hours of bug bounty attacks, Anthropic is inviting the wider public to test out the system to see if it can fool it into breaking its own rules.
In a new paper and accompanying blog post, Anthropic says its new Constitutional Classifier system is spun off from the similar Constitutional AI system that was used to build its Claude model. The system relies at its core on a "constitution" of natural language rules defining broad categories of permitted (e.g., listing common medications) and disallowed (e.g., acquiring restricted chemicals) content for the model.
© Getty Images
DeepSeek's model is censored at both the application and training layers, a Wired investigation shows.
© 2024 TechCrunch. All rights reserved. For personal use only.
Meta CEO Mark Zuckerberg has pledged to make artificial general intelligence (AGI) — which is roughly defined as AI that can accomplish any task a human can — openly available one day. But in a new policy document, Meta suggests that there are certain scenarios in which it may not release a highly capable AI […]
© 2024 TechCrunch. All rights reserved. For personal use only.
Following Thursday’s announcement, Gemini 2.0 Flash is now appearing in the Gemini app on Android and iOS.
more…
Apple Notes gained a variety of great features in iOS 18 last fall, like collapsible sections, color highlighting, audio recording with transcripts, and more. But I recently discovered one more change that answers one of my long-time feature requests, and could satisfy a lot of power users.
more…
Following Thursday’s announcement, Gemini 2.0 Flash is now appearing in the Gemini app on Android and iOS.
more…
US District Court Judge Amit Mehta denied Apple’s emergency request to halt the Google Search monopoly trial that could dismantle their lucrative search that’s reportedly worth as much as $18 billion a year. The order came in late Sunday, with Judge Mehta saying Apple hasn’t demonstrated satisfactory reasons for its emergency motion to stay that was filed on January 30th.
Apple said last week that it needs to be involved in the Google trial because it does not want to lose “the ability to defend its right to reach other arrangements with Google that could benefit millions of users and Apple’s entitlement to compensation for distributing Google search to its users.”
The remedies phase of the trial is set for April, and lawyers for the Department of Justice have argued that Google should be forced to sell Chrome, with a possibility of spinning off Android if necessary. While Google will still appeal the decision, the company’s proposed remedies focus on undoing its licensing deals that bundle apps and services together.
“Because Apple has not satisfied the ‘stringent requirements’ for obtaining the ‘extraordinary relief’ of a stay pending appeal, its motion is denied,” states Judge Mehta’s order. Mehta explains that Apple “has not established a likelihood of success on the merits” for the stay. That includes a lack of clear evidence on how Apple will suffer “certain and great” harm.
Google was found liable for holding an illegal monopoly on general search, and the remedies phase of the trial is set for April, where Google businesses like Android, Chrome, and Search could be broken up.
It looks like Buffy the Vampire Slayer is coming back from the dead (again) as part of a new Hulu series.
Deadline reports that Hulu is close to finalizing a pilot order for a Buffy the Vampire Slayer follow-up set to be directed by Chloé Zhao and written by Nora and Lilla Zuckerman (Poker Face). Dolly Parton is attached as an executive producer for the pilot, while Zhao, the Zuckerman sisters, Gail Berman, Fran Kuzui, and Kaz Kuzui will produce the series as a whole. As Deadline notes, Joss Whedon will not be working on the project.
No details about the show have been announced, but in addition to executive producing, Sarah Michelle Gellar is expected to reprise her role as Buffy Summers. After the original show went off the air in 2003, Buffy’s adventures continued on in the pages of Dark Horse’s comics series that ran until 2018.
Boom! Studios rebooted the franchise in its own 2018 comics series that ended up taking Buffy in a new direction. In terms of monsters and magic, both comics series got bigger than the original show ever did, which might make adapting them more expensive than Hulu is down for. But with so much Buffy lore out in the ether to pull from, the streamer can probably work something out.
Artemiy Pavlov, the founder of a small but mighty music software brand called Sinesvibes, spent more than 15 years building a YouTube channel with all original content to promote his business' products. Over all those years, he never had any issues with YouTube's automated content removal system—until Monday, when YouTube, without issuing a single warning, abruptly deleted his entire channel.
"What a 'nice' way to start a week!" Pavlov posted on Bluesky. "Our channel on YouTube has been deleted due to 'spam and deceptive policies.' Which is the biggest WTF moment in our brand's history on social platforms. We have only posted demos of our own original products, never anything else...."
Officially, YouTube told Pavlov that his channel violated YouTube's "spam, deceptive practices, and scam policy," but Pavlov could think of no videos that might be labeled as violative.
© Talaj | iStock / Getty Images Plus
The Republican Senator from Missouri Josh Hawley has introduced a new bill that would make it illegal to import or export artificial intelligence products to and from China, meaning someone who knowingly downloads a Chinese developed AI model like the now immensely popular DeepSeek could face up to 20 years in jail, a million dollar fine, or both, should such a law pass.
Kevin Bankston, a senior advisor on AI governance at the Center for Democracy & Technology, told 404 Media it is “a broad attack on the very idea of scientific dialogue and technology exchange with China around AI, with potentially ruinous penalties for AI researchers and users alike and deeply troubling implications for the future of online speech and freedom of scientific inquiry.”
Hawley introduced the legislation, titled the Decoupling America’s Artificial Intelligence Capabilities from China Act, on Wednesday of last year.
“Every dollar and gig of data that flows into Chinese AI are dollars and data that will ultimately be used against the United States,” Senator Hawley said in a statement. “America cannot afford to empower our greatest adversary at the expense of our own strength. Ensuring American economic superiority means cutting China off from American ingenuity and halting the subsidization of CCP innovation.”
Hawley’s statement explicitly says that he introduced the legislation because of the release of DeepSeek, an advanced AI model that’s competitive with its American counterparts, and which its developers claimed was made for a fraction of the cost and without access to as many and as advanced of chips, though these claims are unverified. Hawley’s statement called DeepSeek “a data-harvesting, low-cost AI model that sparked international concern and sent American technology stocks plummeting.”
Hawley’s statement says the goal of the bill is to “prohibit the import from or export to China of artificial intelligence technology, “prohibit American companies from conducting AI research in China or in cooperation with Chinese companies,” and “Prohibit U.S. companies from investing money in Chinese AI development.”
Hawley’s bill and its aims were covered credulously on Fox News, but even if you think the bill’s goals are worth pursuing the actual language of the bill is broad and dystopian. Unlike legislators who fearmongered about TikTok and wanted to ban it, Hawley’s bill would criminalize the activity of average users, millions of whom downloaded DeepSeek recently, making it one of the most popular apps on the Apple App store.
Specifically, the bill prohibits “the importation into the United States of artificial intelligence or generative artificial intelligence technology or intellectual proprietary developed or produced in the People’s Republic of China.” Those who violate this “Shall be subject to the criminal penalties set forth in subsection (b) of section 1760 of the Export Control Reform Act of 2018 (50 U. S.C, 4819).”
That law states that “A person who willfully commits, willfully attempts to commit, or willfully conspires to commit, or aids and abets in the commission of, an unlawful act described in subsection (a) (1) shall be fined not more than $1,000,000; and (2) in the case of the individual, shall be imprisoned for not more than 20 years, or both.”
the Center for Democracy & Technology’s Bankston told me that he’s skeptical that there would be strong criminal cases against someone who unintentionally downloaded an app like DeepSeek because the legislation specifies a person’s conduct must be “willful” for the imposition of criminal penalties, the bill is still “worrisomely broad.”
“It appears that it *could* apply to someone who downloaded DeepSeek knowing that it was from China, and yes, the criminal penalty for that under this proposal would be up to one million dollars or 20 years in prison (and also potentially civil penalties as well, which may require less proof of state of mind and may potentially even reach a mere accidental ‘importer’ of a Chinese model),” Bankston said.
The bill, which also prohibits the “transfer of research,” could create an unworkable environment for computer scientists who make their research public, and regularly read AI papers published by Chinese researchers.
“Beyond just impacting people downloading models from China, the bill's penalties for the import to or export from China of AI technology and intellectual property could also potentially extend to anyone who publishes AI models or research papers on the open internet knowing they will be downloaded by people in China,” Bankston said. “Researchers are also threatened by the second half of the bill, which would directly outlaw American collaboration with researchers at basically any Chinese university or company—with a fine of up to 100 million dollars for any company that violates the prohibition, amongst other penalties.”
"The bill threatens the development and publishing of AI advancements in the United States, and we're particularly worried about the impact on open and collaborative development of these technologies outside the proprietary systems of the Big Tech incumbents," Kit Walsh, the Electronic Frontier Foundation's Director of AI and Access-to-Knowledge Legal Projects, told me. "In the past, the government has argued that merely publishing information on the internet counts as an export, and interpreting this law in such a way would further solidify the dominance of proprietary AI over open or academic research. The law would also interfere with efforts at AI accountability, such as transparency requirements that states and members of Congress have sought to create in order to make sure that AI isn't harming people in the United States when used for decisions about such wide-ranging things as housing, health care, and hiring."
On its face, the bill seems mostly like hawkish posturing from Hawley, and the language of the bill seems unworkable given the current state of computer science, the AI industry, and the culture of researchers sharing their work. However, there is bipartisan support for legislation that targets China wherever it appears able to topple American dominance. Banning TikTok also seemed like a ludicrous notion at first given its popularity among Americans, and while the app is still live, a bill banning it did pass both the house and the Senate and was signed by the president.
Hawley’s office did not respond to a request for comment.
Update: This article has been updates with comment from the EFF.
Ahead of its launch in March, Nothing is teasing a dedicated camera button for its next smartphone, the Nothing Phone (3a).
more…
Meta isn’t shy about its ambitions to make its Ray-Ban smart glasses a mainstream gadget. So much so, it’s made not one but two Super Bowl ads featuring not one, but two Marvel actors. (And Kris Jenner.) And as a little cherry on top, Meta announced it’s creating a limited edition Super Bowl version of the glasses that won’t be for sale to the general public.
You can already watch the first of the two Super Bowl ads, which features Chris Hemsworth and Chris Pratt causing chaos in Kris Jenner’s personal art collection. The actors are shown asking the Meta AI to identify art pieces, as well as translate foreign languages. Jenner, meanwhile, is shown asking the AI glasses to ring up her lawyer after an incident with a $6.2 million banana.
In a blog about the ads, Meta says the limited edition glasses will be a pair of matte black Wayfarers with either gold, mirror-like lenses or lenses featuring the team colors of the Super Bowl finalists. The glasses will also come with a custom-etched case. However, they won’t be available for purchase, hinting that these will likely be seeded to celebrities and influencers to drum up buzz.
There are a couple things at play here. For starters, smart glasses have come a long way with the public from the debacle that was Google Glass. Meta announced last week that the Ray-Bans have sold over 1 million units in 2024. That’s the type of success for an emerging gadget category that could justify an expensive Super Bowl ad. Second, Meta has recently shown it’s keen on pushing that success further using new styles and limited edition versions, not unlike how Nike or other sneaker companies create hype. Meta will purportedly release Oakley-branded smart glasses later this year for athletes, and last year, it also had a limited-edition translucent model of the Ray-Bans for its Meta Connect event.
Nuance Audio, new "hearing glasses" from EssilorLuxottica, have received FDA clearance and "a CE marking under the Medical Devices regulation in the EU," making them officially ready to sell in the US and Europe.
The smart glasses are essentially prescription frames with an over-the-counter (OTC) hearing aid built-in. Using open-ear speakers and a beamforming microphone array, Nuance Audio can raise the volume of whoever you're speaking to, while attempting to cancel out whatever noise is going on around you. You can control the glasses with a companion app or a dedicated remote.
The glasses come in only two shapes, two colors and three sizes, far fewer options than EssilorLuxottica's traditional glasses brands, or the Ray-Ban Meta smart glasses, but the tech seems like it could be ported to new frames down the road. Engadget deputy editor Cherlynn Low was able to try on Nuance Audio during CES 2025 and found them to be very comfortable and lightweight, though she wasn't able to adequately test their capabilities as a hearing device.
Nuance Audio is just the latest in a wave of new hearing aid-like devices that have been released since the FDA introduced its ruling on OTC hearing aids in 2022. Similar to Nuance Audio, Apple's AirPods Pro 2 now offer software-enabled hearing aid functionality, along with a built-in hearing test and a hearing protection feature for loud environments. While turning every pair of AirPods Pro 2 into a hearing aids is great from a cost-saving perspective, getting a pair of Nuance Audio frames that can also double as your prescription glasses seems potentially even more convenient.
EssilorLuxottica says Nuance Audio will be available in the US in Q1. "Certain European countries" will also get the glasses in the first half of 2025, starting with Italy in Q1, and then France, Germany and the UK. The company hasn't shared official pricing for the Nuance Audio.
This article originally appeared on Engadget at https://www.engadget.com/wearables/the-fda-and-eu-clear-nuance-audio-smart-glasses-with-built-in-hearing-aids-for-sale-211053856.html?src=rss©
© EssilorLuxottica
Uber rolled out a new set of features on Monday that could remove some friction for people with service animals. The company now lets handlers automatically notify drivers that they ride with a service animal.
In a blog post, Chris Yoon, an Uber Product Manager, wrote that he “joined Uber with a singular mission: to help expand accessibility in rideshare, including for those who rely on service animals.” He says he and other blind travelers often message drivers in advance to let them know about their service animals. Today’s new features can automate that for them if they choose.
Drivers who consider refusing rides will have to think twice — and then some. If the driver requests a cancelation after learning about the service animal, Uber will automatically send an in-app reminder that doing so is against not only company policy but also the law. Uber says drivers refusing rides with service animals “may permanently lose access to the platform.”
Passengers who opt into the new features will automatically receive a message after a driver cancels, asking what happened and offering additional support. Those who don’t use the feature can contact the company through the app or over the phone.
Riders can set up the feature through the Uber app. Go to Account > Settings > Accessibility, and tap Service Animal. It will then prompt you to fill out an eligibility form and pick which parts of the feature (like notifying the driver in advance) you want to turn on.
This article originally appeared on Engadget at https://www.engadget.com/transportation/uber-wants-to-make-rides-easier-for-people-with-service-animals-210958456.html?src=rss©
© Uber
