AI News Roundup – Los Angeles ramps up use of AI-assisted cameras to issue traffic tickets, Google launches AI-based language learning tools, Generative AI in video games, and more

To help you stay on top of the latest news, our AI practice group has compiled a roundup of the developments we are following.

• • The Los Angeles Times reports that Metro, the public transit agency of the California city, has issued nearly ten thousand traffic citations using AI-assisted cameras. The cameras were first installed on some Metro buses last year, and automated ticketing began in February of this year. Each camera scans for vehicles illegally parked in bus lanes or in front of bus stops using AI identification. If a violation is detected, the camera compiles a video of the incident, a picture of the vehicle’s license plate, and other information which is then reviewed by a human at Metro before the citation is sent to the vehicle’s owner. Such citations carry a minimum fine of $293, and the system has identified thousands of violations over its first few months in operation. In the first two weeks of April, over 6,600 citations were issued, in comparison to around 570 per month issued by police officers before the camera system was deployed. L.A. County Supervisor Janice Hahn said that “a single parked car in a bus lane can delay dozens or even hundreds of riders. So, this is not just about issuing tickets, it’s about making sure our Metro buses run on schedule and get our riders to their destinations on time.” Similar systems have been proposed or used in other major American cities, including New York, Chicago and Washington, D.C. 
• • Inc reports on Google’s launch of three experimental AI-powered language learning tools called “Little Language Lessons.” The first, “Tiny Lessons,” is designed to teach words and phrases in a given language using specific scenarios, such as going on a first date or taking a taxi. While the scenarios are limited in the current release of the tools, further AI-generated situations may be available in consultation with language educators and linguists. The second tool, “Slang Hang,” is designed to teach users slang terms, idioms and other phrases in a target language through AI-generated conversations between two people and offering translations and explanations of terms used. The third tool, “Word Cam,” allows users to point their device camera at an object in the world and be given the word for that object in a target language. While each tool is limited in scope and is intended as a demonstration of the abilities of Google’s Gemini AI model, the company told Inc that it does not plan to expand its offerings, which would put it in competition with other language learning apps such as Duolingo. However, related language-related AI features may be in the works at Google, as a breakdown of the latest update to Google Translate’s Android app by Android Authority found an unfinished “Practice” feature that could allow users to use activities to help learn their target language. 
• • The Financial Times analyzes the impact of generative AI on the video game industry. Video games have often been on the cutting edge of technological advancement, and the term “AI” itself has been used in the context of games long before generative AI exploded onto the scene in 2022. Now dubbed “classical AI,” this type refers to the rules that non-player characters or enemies follow in the game’s virtual environment. Generative AI, on the other hand, has a variety of applications in the industry, such as creating assets used in development (such as 3-D models), generating code and story text, and generating voices without the need for voice actors. Microsoft and Google have each used video games in their generative AI research, with the former creating a model that can simulate a game using gameplay footage, while the latter’s Genie model can generate an entire 3-D virtual world from an image prompt. Beyond these research projects, however, the use of generative AI has been limited in mainstream games, though smaller studios have introduced it in small doses. Larger players in the industry have used AI in different ways: Microsoft uses machine learning to help detect cheaters in some online games, while Activision has used generative AI for some in-game assets. Other companies are more skeptical — Japan’s Nintendo has said it will go in a “different direction” than other companies and not use AI in its games, while the CEO of Take-Two (the publisher of “Grand Theft Auto”) called artificial intelligence “an oxymoron.” The larger effect of generative AI on the industry has yet to be seen, partly due to the complexity of game development and pushback on its use among workers in the field. 
• • The Washington Post reports on the U.S. launch of World, a startup that creates digital IDs based on iris scanning. World, formerly known as WorldCoin and backed by Sam Altman, the CEO of OpenAI, was inspired by the need for a reliable way to distinguish people from AI systems online. Iris-based identification is not a new concept, but World has developed an orb-shaped device that takes images of a person’s face and eyes and then creates an encrypted, anonymized World ID. Those who create a World ID receive 150 tokens of the company’s Worldcoin cryptocurrency, each of which is currently worth just over one U.S. dollar. World previously had scanned irises in countries such as Argentina and Kenya, but the latter’s government shut down its activities over concerns regarding its privacy and data security practices. World has partnered with Visa, the dating app company Match, and the payment system Stripe to provide verification services, though it remains to be seen how widespread the company’s “proof-of-human” ID will become in the coming months in the U.S. 
• • The Associated Press reports that Meta this past week hosted its first-ever “Llamacon,” a developer conference named for its flagship line of AI models. The company announced several new AI features and products at the event, including a standalone AI app based on Llama 4, aimed to directly compete with rival OpenAI’s ChatGPT. Previously, Meta AI had only been integrated into Meta’s social media apps Facebook, Instagram, and WhatsApp, though the new standalone app will also have a social media component that allows users to see how others are using the AI system. The company also announced an application programming interface (API) for its Llama models, saying that “the models you build on the Llama API are yours to take with you wherever you want to host them, and we don’t keep them locked on our servers.”