Tech Brew Ride Home - Fri. 06/21 – Kaspersky Banned By The US Government

Episode Date: June 21, 2024

The government has banned Kaspersky antivirus sales in the US. People are losing their minds over Claude 3.5 Sonnet from Anthropic. Soon all devices can pair to your iPhone as easily as AirPods do. And, of course, the Weekend Longreads Suggestions. Links: US bans sale of Kaspersky software citing security risk from Russia (TechCrunch) Anthropic claims its latest model is best-in-class (TechCrunch) Anthropic’s Claude 3.5 Sonnet wows AI power users: ‘this is wild’ (VentureBeat) New iOS 18 API brings AirPods setup experience to third-party accessories (9to5Mac) SpaceX unveils backpack-sized ‘Starlink Mini’ satellite internet antenna for $599 (CNBC) Weekend Longreads Suggestions: Why are mobile game ads so weird and bad? (Sherwood) From Infocom to 80 Days: An oral history of text games and interactive fiction (ArsTechnica) The song Stevie Nicks wrote to “haunt” Lindsey Buckingham (Far Out) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech meme right home for Friday, June 21st, 2024. I'm Brian McCullough today. The government has banned Kaspersky anti-virus sales in the U.S. People are losing their minds over Claude 3.5 Sonnet from Anthropic. Soon, all devices compare to your iPhone as easily as AirPods do. And of course, the weekend long read suggestions. Here's what you missed today in the world of tech. The U.S. Commerce Department has banned Kaspersky anti-virus sales in the United States, saying Kaspersky threatens national security. and user's privacy because it is based in Russia. Quoting TechCrunch. Russia has shown it has the capacity, and even more than that, the intent to exploit Russian companies like Kaspersky to collect and weaponize

the personal information of Americans. And that's why we are compelled to take the action that we're taking today, U.S. Commerce Secretary Gina Raimondo said in a call with reporters. Kaspersky will be banned from selling its software to American consumers and businesses starting on July 20th, but the company will be able to provide software and security updates to existing customers until September 29th. After that, Kaspersky will no longer be permitted to push software updates to U.S. customers, according to Raimondo. That means your software and services will degrade. That's why I strongly recommend that you immediately find an alternative to Kaspersky,

Ramando said. Raimondo said that U.S. consumers who already used Kaspersky's antivirus are not violating the law. U.S. individuals and businesses that continue to use or have existing Kaspers, Fareski products and services are not in violation of the law. You have done nothing wrong, and you are not subject to any criminal or civil penalties, said Ramando. However, I would encourage you in the strongest possible terms to immediately stop using that software and switch to an alternative in order to protect yourself and your data and your family, end quote. To inform consumers, Ramando said the Department of Homeland Security

and the Justice Department will work to notify U.S. consumers and the U.S. government will set up a website, quote, so people who are impacted can find the information they need to understand why we're doing what we're doing and help them take next steps, end quote. A senior U.S. Commerce Department officials said during the press call that federal cybersecurity agency, CISA, will do outreach to critical infrastructure organizations that use Kasperski software in their operations to help them find alternatives. The official also said that they don't plan on naming any specific action by Kaspersky that led to today's decision.

The Commerce Department asked reporters not to do that. to name the official. The ban announced Thursday is the latest escalation and a long series of U.S. government actions against the Moscow headquartered Kaspersky. In September 2017, the Trump administration banned U.S. federal agencies from using Kaspersky software over fears that the company could be compelled to help Russian intelligence agencies. Earlier in the year, it was reported that Russian government hackers had stolen U.S. classified documents stored on an intelligence contractor's home computer because it was running Kaspersky's antivirus, making the first known incident of espionage resulting from use of the company's software. The decision to ban Kasperski

has been in the work since last year, according to a report by the Wall Street Journal in April of 2003. According to Kaspersky, the company has more than 400 million individual customers and over 240,000 corporate customers worldwide. The senior official declined to say how many U.S. customers Kaspersky has, but said there is a significant number, including critical infrastructure organizations and state and local government entities, end quote. So again, to be clear, The restrictions on Kaspersky will stop all new sales of its antivirus software in the U.S. beginning July 20th, but will also bar software updates to U.S. customers after September 29th. So, act accordingly if you're a Kaspersky customer.

Yesterday, Anthropic launched Claude 3.5 Sonnet, which it says beats its flagship model Claude 3 opus and outperforms GPT40 in some tests. It's available for free on the web and iOS right now. Coding TechCrunch. Claude 3.5 Sonnet can analyze both text and images as well as generate text and its Anthropics best-performing model yet, at least on paper. Across several AI benchmarks for reading, coding, math, and vision, Claude 3. Sonnet outperforms the model. It's replacing Claude 3 Sonnet and beats Anthropics' previous flagship model Claude 3 Opus.

Benchmarks aren't necessarily the most useful measure of AI progress, in part because many of them test for esoteric edge, that aren't applicable to the average person, like answering health exam questions. But for what it's worth, Claude 3.5 Sonnet just barely best rival leading models, including OpenAIs recently launched GPT40 on some of the benchmarks Anthropic tested it against. Alongside the new model, Anthropic is releasing what it's calling artifacts, a workspace where users can edit and add to content, e.g. code and documents generated by Anthropics models. Currently in preview, artifacts will gain new features like ways to collaborate with larger teams and store knowledge bases in the near future, Anthropic says. Claude 3.5 Sonnet is a bit more performant than Cloud

3 Opus, and Anthropics says that the model better understands nuanced and complex instructions in addition to concepts like humor. But perhaps more importantly for devs building apps with Claude that require prompt responses, e.g. Customer service chatbots, Cloud 3.5 Sonnet is faster. It's around twice the speed of Cloud 3 Opus Anthropic claims. Vision. Analyzing photos is one area where Claude 3.5 Sonic greatly improves over three opus, according to Anthropic. Cloud 3.5 Sonic can interpret charts and graphs more accurately and transcribe texts from imperfect images, such as picks with distortions and visual artifacts. Michael Gersten Haber, product lead at Anthropics, says that the improvements are the result of

architectural tweaks and new training data, including AI-generated data, which data specifically Girsten Haber wouldn't disclose, but he implied that Claude 3.5 Sonnet draws much of its strength from these training sets. So what's the significance of Claude 3.5 Sonnet in the broader context of Anthropic and the AI ecosystem itself, for that matter? Claude 3.5 Sonnet shows that incremental progress is the extent of what we can expect right now on the model front, barring a major research breakthrough. The past few months have seen flagship releases from Google, Gemini 1.5 Pro, and OpenAI, GPT40. that move the needle marginally in terms of benchmark and qualitative performance,

but there hasn't been a leap of matching the leap from GPT3 to GPT4, for instance, in quite some time, owing to the rigidity of today's model architectures and the immense compute they require to train. As generative AI vendors turn their attention to data curation and licensing, in lieu of promising new scalable architectures, there are signs investors are becoming wary of the longer-than-anticipated path to ROI for generative AI. Anthropic is somewhat inculcated from this pressure, being in the enviable position of Amazon's and, to a lesser extent, Google's insurance against OpenAI. But the company's revenue forecasted to reach just under $1 billion by year-end 2024 is a fraction of open AIs, and I'm sure Anthropics backers don't let it forget that fact. Despite a growing customer base that includes household brands such as Bridgewater, Brave, Slack, and DuckDuck Go, Anthropic still lacks a certain enterprise cachet.

tellingly, it was Open AI, not Anthropic, with which PWC recently partnered to resell generative AI offerings to the enterprise. So Anthropic is taking a strategic and well-trodden approach to making inroads, investing development time into products like Cloud 3 Sonnet to deliver slightly better performance at commodity prices. Cloud 3.5 Sonnet is priced the same as Cloud 3 Sonnet, $3 per million tokens fed into the model, and 15 per million tokens generated by the model, end quote. Now, as you heard, that was a bit hedgy, but I have to say that online, the response has been quite the opposite, almost nothing short of euphoric. In the show notes I've linked to a venture beat post collecting people's reactions and some of the things that they've already developed with it.

The consensus, at least online, seems to be that Sonnet appears to be a tremendous leap for Anthraic, and LLMs generally may be showing that AI model makers' performance gains are not slowing down. iOS 18 is apparently bringing a new API called Accessory Setup Kit, which will let third-party accessory makers build the same quick-paring experience as Apple accessories like AirPods have, quoting 9-5 Mac. When Apple introduced AirPods in 2016, the company also unveiled a new, easy, and intuitive way to pair wireless accessories to iPhone and iPad. Rather than having to go to Bluetooth settings and press buttons, the system identifies the accessory nearby and prompts the user to pair it. With iOS 18, this quick pairing process will be available for the first time to accessory makers. With just a tap,

the system will automatically handle all the Bluetooth or Wi-Fi connectivity required by the accessory. This also means that users will no longer have to manually give Bluetooth and Wi-Fi permissions individually to that accessory's app. If the accessory requires a more complex pairing process, such as confirming a PIN code, the iOS 18 API can also ask the user for this information without the need to open an app. Once the accessory has been paired, more information about it can be found in a new accessories menu within the privacy settings. This is certainly huge news for accessory makers who had no chance of offering the same level of integration that Apple accessories have with iOS. Of course, this will require updates to both accessory apps and the accessories themselves,

so don't expect to see the new API in action before the release of iOS 18 this fall, end quote. SpaceX has unveiled Starlink Mini, a compact version of its satellite internet antenna, offering a limited number for $599 each, quoting CNBC. Starlink Mini is a compact portable kit that can easily fit in a backpack designed to provide high-speed, low-latency internet on the go, according to a customer email sent by SpaceX on Wednesday and viewed by CNBC. The company is offering a limited number of the Starlink Mini Antennas for $599 each in an early access release.

That's $100 more than the base model standard antenna sold with its residential service, although the company aspires to reduce the price tag. Our goal is to reduce the price of Starlink and especially for those around the world where connectivity has been unaffordable or completely unavailable, SpaceX wrote in the email. In addition to the upfront hardware cost, service for a Starlink Mini is effectively $150 per month, as SpaceX is offering the service for a Mini as an additional $30 per month bundle on top of a $120 per month residential service. The Mini Roam Service can be used anywhere in the United States but has a cap of 50 gigabytes of data per month with Starlink charging $1 per gigabyte for additional data. The Starlink Mini antenna is about the size and weight of a laptop

at just over two pounds, and measuring about 12 inches by 10 inches by 1.5 inches. It's roughly half the size and one-third the weight of Starlink's standard antenna. SpaceX's email said Starlink Mini comes with a built-in Wi-Fi router and, quote, lower power consumption than its other antennas, yet it still boasts download speeds of over 100 megabits per second. The email did not specify when Starlink Mini deliveries would begin in a post on social media, Vice President of Starlink Engineering Michael Nichols said the company is ramping production on Starlink Mini and that it, quote, will be available in international market soon. SpaceX CEO Elon Musk wrote in a post Monday that setting up a Starlink Mini took less than five

minutes. This product will change the world, Musk declared, end quote. Time for the weekend long-rate suggestions. First up from Sherwood, why do you see those mobile game ads for Royal Match where you have to save the King in like every other ad everywhere? Quote, Killing the King has been a core part of the Royal Match advertising strategy since Dream Games first released the app in 2021. It's an odd approach. McDonald's probably wouldn't drown Grimmis if you don't buy a hamburger, but it seems to be working. In April, Royal Match surpassed $3 billion in lifetime user spending just five months after hitting 2 billion. million dollars. For the supposed three billion people around the world who play these games,

sometimes until our phones are hot enough to cook an egg, these unsettling ads are part of life. Popping up on Instagram, X, and YouTube, and in between levels of other games, mobile game ads are a key part of a massive business. Royal Match ads fit into a category I call someone's playing a game wrong. The ads will feature a clip of a word game, shooter game, or puzzle, but the person controlling the gameplay is a masochist. In a 30-second ad for a word on scrambling game, they'll repeatedly attempt something like hitnik, but never think. In a shooting game where players control a mob that moves through barriers that change in size, the ads player will frequently select a counterintuitive option like subtract 10 instead of add 50.

Stupid stuff, but the psychology is pretty clear. Create an ad so unsatisfying with so obvious an answer that someone might be just outraged enough to download the game and try to beat it themselves. The strategy has ruffled some feathers. A Reddit community dedicated to the niche industry R-slash-Shitty mobile game ads has almost 160,000 subscribers, end quote. Next, Ars Technica takes a look at the history of text-based gaming and the interactive fiction space in general. Quote, the interactive fiction story began in the 1970s. The first widely played game in the genre was colossal cave adventure, also known simply as adventure. The text game was made by Will Crowther in, in 1976, based on his experiences splunking in Kentucky's aptly named Mammoth Cave.

Descriptions of the different spaces would appear on the terminal, then players would type in two-word commands, a verb followed by a noun, to solve puzzles and navigate the sprawling in-game caverns. During the 1970s, getting the chance to interact with a computer was a rare and special thing for most people. The roots of interactive fiction are entangled with the roots of the computing industry. I think it's always been a focus on the written word as an engine for what we consider a game, said software developer and tech entrepreneur Lisa Daly. Originally, that was born out of necessity of primitive computers of the 70s and 80s,

but people discovered that there was a lot to mind there, end quote. Home computers were just beginning to gain traction as Stanford University student Don Woods released his own version of adventure in 1977 based on Crowther's original Fortran work without wider access to comparatively pint-sized machines like the Apple 2 and the VIC-20. Scott Adams might not have found an audio. for his own tech adventure games, released under his company Adventure International, and another homage to Crowther. As computers spread to more people around the world, interactive fiction was able to reach more and more readers, end quote. And finally, a piece from Far Out magazine.

It's not tech, but I've become obsessed with the story of the Fleetwood Mac song Silver Springs. It's the song Stevie Nix wrote to haunt Lindsay Buckingham about the fact that he had dumped her. quote, one song that didn't make the final cut for the album Rumors is Silver Springs, which was replaced on the record's final cut by I Don't Want to Know, much to the frustration of Nix. On the track, she tries to seek revenge upon Buckingham to haunt her former bow. In Silver Springs, Nick's sings, I know I could have loved you, but you would not let me.

I'll follow you down till the sound of my voice will haunt you. In a 2009 interview with Rolling Stone, Nix revealed she wanted Buckingham to suffer the same pain as her. her. She said, quote, it was me realizing that Lindsay was going to haunt me for the rest of my life, and he has. The song was eventually released as the B-side for Go Your Own Way. And even though they are no longer on speaking terms, Silver Springs is concrete proof of how much Buckingham meant to Nix once upon a time, end quote. Yes, but what he also means to her now? Because, okay, Brian, neat story about a disc track, who cares? But here's the thing. At the bottom of this piece, you'll see a YouTube

video of a live performance of the song from, I don't know, like 20 years ago maybe, this is what I become obsessed with. If you do nothing else, play the video and skip to about the four-minute mark. Stevie and Lindsay are standing next to each other on stage singing the duet of the chorus for Silver Springs. And then Stevie turns and starts singing the song directly in Lindsay's face. She's putting her all into it and singing the words with all her heart. You'll never get away, never get away. My voice will haunt you, the woman who loved you. To his credit, Lindsay does not look away, but also see the pained look on his face. He's like, I know what you're doing here, I know what this is, I just have to stand here and take this.

It's one of the most powerful live performances I've ever seen, the pain and the anger in her voice, the pain and chagrin on his face. It's amazing. No weekend bonus episode for you. This weekend, talk to you on Monday.