| |
| Here's a look at today's AI briefing: - OpenAI could release GPT-5 this year, Insider reports.
- EU could pass rules to fight AI misinformation during elections.
- France fines Google $270M for breaching AI deal commitments with news publishers.
- Judge declines to sanction Michael Cohen over fake AI legal citations.
- Three researchers behind Stability AI's Stable Diffusion are leaving.
Beth
p/beth-duckett | |
| 1 | | OpenAI could potentially release GPT-5 around mid-year, possibly in the summer, according to sources cited by Business Insider. Some enterprise customers have been shown demos of the newest model and its improvements to ChatGPT, according to a source. More: - One CEO, who was not named, told Insider that GPT-5 is "really good, like materially better" than GPT-4.
- The CEO said OpenAI showcased the new model's specific applications for his company.
- He hinted at future features, including GPT-5's ability to summon AI agents to autonomously perform tasks.
- Another source said OpenAI is finalizing GPT-5's training before doing internal safety tests and external "red teaming."
- The testing could delay the model's release, and without a set launch date, internal timelines could change, Insider noted.
Zoom out: - During a new podcast interview with Lex Fridman, CEO Sam Altman said OpenAI will "release an amazing new model this year," without confirming if it's GPT-5.
- "I think before we talk about a GPT-5-like model, we have a lot of other important things to release first," he said.
- Altman also admitted that he thinks GPT-4 "kind of sucks" before adding that the jump from GPT-4 to GPT-5 would mirror the improvements seen from GPT-3 to GPT-4.
|     | |
| 2 | | European lawmakers are poised to pass new rules on major online platforms to counter AI-generated disinformation and deepfakes during elections, according to the Financial Times. The rules would fine tech companies up to 6% of global turnover if they fail to comply. More: - The European Commission could approve the rules as early as next week to address false content, manipulation, and deepfakes before the European elections in June.
- Platforms would be required to form teams to monitor disinformation in 23 languages and collaborate with EU cybersecurity agents across member states.
- The regulations would be enforced under the EU's Digital Services Act (DSA), targeting major platforms and search engines like Google, Facebook, and Amazon.
Zoom out: - Last month, 20 tech companies, including OpenAI, Amazon, and Google, signed an accord aimed at stopping AI from misleading voters and undermining elections worldwide.
- The companies agreed to work together on tools to detect deceptive AI-generated media, along with voter education campaigns, plus watermarking and metadata embedding.
- At least 83 elections are scheduled around the world this year, the highest concentration expected in the next 24 years.
| | | |
| A message from our sponsor, MM01 AI LIMITED. |  | | Unlocking the future of AI Agents - $1 Million Challenge With a prize pool of $1 million, this challenge stands as one of the most prestigious and generous in the AI arena, open to attract the brightest minds from around the globe. Whether you're an AI enthusiast, a pioneering innovator, a student, or a professional, this is your arena to shine. In this challenge, creativity meets complexity, and intelligent solutions are born for real-world impact. Unleash your creativity and expertise across a spectrum of fields, from healthcare to cybersecurity, and beyond. Push AI Agents beyond the frontiers of today’s capabilities. Apply Today | |
|
| 3 | | Google was fined €250M ($270M) by France's competition watchdog, in part for training its Bard AI model on publishers' news content without their consent. French officials accused Google of breaking its commitments by not allowing news publishers to opt out of training Bard, now called Gemini. More: - French authorities accused Google of reneging on deal commitments with local news outlets by using their content to train Bard without permission.
- France's ADLC found that Google failed to fulfill commitments to publishers, including by not informing them or the ADLC about using their content for Bard training and outputs and delaying opt-out options until 2023, hindering fair negotiations.
- Google called the fine excessive but accepted the settlement to "move on" and focus on sustainable content strategies and constructive partnerships with French publishers.
- The company has agreed not to dispute the facts and proposed remedies as part of the settlement, according to the watchdog.
Zoom out: 
| | | |
| 4 | | A federal judge declined to sanction Michael Cohen for mistakenly submitting fake legal citations created by AI, though the judge still described the incident as "embarrassing and certainly negligent." Cohen, Donald Trump's former fixer, mistakenly used fake legal citations from Google's Bard in a motion filed last year. More: - Cohen gave his lawyer, David Schwartz, the fake citations to three nonexistent cases for the motion filed in Federal District Court.
- Cohen, a disbarred attorney, said he mistakenly thought Google Bard was a "supercharged search engine," not realizing it generates text and can create citations and descriptions that appear real "but actually were not."
- Cohen and Schwartz later apologized. In his ruling, U.S. District Judge Jesse Furman in Manhattan called Schwartz's conduct "certainly negligent, perhaps even grossly negligent," but saw no malintent to warrant sanctions.
- "As embarrassing as this unfortunate episode was for Schwartz, if not Cohen, the record does not support the imposition of sanctions," Furman wrote.
Zoom out: - Furman also declined to end the court-supervised release for Cohen, who pleaded guilty to tax evasion and campaign finance violations in 2018 and served time in prison.
- The ruling coincides with Cohen's preparations to testify at Trump's upcoming New York criminal trial.
| | | |
| A message from our sponsor, Ritual. |  | | The juggle is real. Find your footing with Ritual Stress Relief.* In 2022, a study conducted by the American Psychological Association reported that 76% of participants reported experiencing health impacts due to stress in the last month. Balancing responsibilities from work, kids, and finances can feel overwhelming and trigger increased cortisol levels. This can throw you off balance, leaving you in an activated “always on” state that’s not physiologically or mentally beneficial. Ritual’s Stress Relief supplement is formulated with 3 clinically studied ingredients combined with their unique BioSeriesTM technology to help support the body’s natural cortisol response for all-day support (8-hour release).* - Shoden® Ashwagandha supports normal cortisol levels to help reduce stress*
- Suntheanine® L-Theanine provides quick stress relief within the hour and a non-drowsy sense of calm*
- Saffron® as affron® promotes a positive mood and reduces stress*
Try Stress Relief today and get 25% off your first month with code INSIDE25. Try it now
*These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease. | |
|
| 5 | | Three of the five researchers behind Stability AI's Stable Diffusion technology are leaving the company, according to Forbes. Stability confirmed the departure of research scientist Robin Rombach, one of the Ludwig Maximilian University of Munich academics who helped develop the popular text-to-image generator. More: - Since 2022, Stability has released its AI image-generation models, starting with Stable Diffusion 1.4 and, most recently, Stable Diffusion 3.0 in February.
- The company has lost employees in the past year. Stability AI replaced its CTO and lost VPs of product, engineering, and R&D, plus a research lead and two LLM leads.
- Last fall, Ed Newton-Rex resigned as leader of Stability's generative AI audio team. His departure was due to a clash over copyright issues, particularly his concerns about AI companies' widespread reliance on the fair use doctrine.
Zoom out: - Despite significant fundraising, the company has faced financial challenges with dwindling cash reserves as costs for computing power far outpace revenue
- Forbes previously reported that researchers affiliated with Stable Diffusion have accused Stability of overstating its role in developing the tool, noting that while the company funded training costs, it didn't contribute to the model's technical creation.
| | | |
| 6 | | 🍋 FRESH FUNDING 🍋 - AI-powered legal contract and negotiation platform DraftWise bagged $20M in Series A funding led by Index Ventures. Existing investors Y Combinator and Earlybird Digital East Ventures participated in the round.
- Hippocratic AI, a startup building safety-focussed LLM for healthcare, closed its Series A funding round at $53M. The firm was valued at $500M. General Catalyst and Premji Invest led the round, joined by Andreessen Horowitz and SV Angels.
- AI-enabled personalized surgery medtech firm Carlsmed raised $52.5M in Series C funding co-led by B Capital and U.S. Venture Partners.
- AI-powered healthtech startup HiLabs secured $39M in Series B funding led by Denali Growth Partners and Eight Roads Ventures.
- Round13 Capital led a $33M Series B funding round into PocketHealth Inc., a startup exploring the use of AI in healthcare. Deloitte Ventures and Samsung Next participated in the round along with existing investors Radical Ventures and Questa Capital.
- AI-powered legal contract and negotiation platform DraftWise bagged $20M in Series A funding led by Index Ventures. Existing investors Y Combinator and Earlybird Digital East Ventures participated in the round.
For the latest on AI startup funding and more in the VC community, subscribe to our Inside Venture Capital newsletter. | | | |
| 7 | | QUICK HITS - Sign up for real-time, easy-to-understand business credit alerts and get the insights you need to help you reach your business goals.*
- Micron shares surged on Wednesday after it forecasted higher-than-expected Q3 revenue and reported an unexpected profit fueled by strong demand for its high bandwidth memory chips used in AI development.
- Microsoft is expected to unveil its first "AI PCs" — the Surface Pro 10 and a Surface Laptop 6 — along with the latest AI updates for Windows 11 and Copilot during an online event tomorrow, March 21.
- AI-driven cybercrime threatens to increase tax season fraud this year, with scammers using AI techniques to impersonate taxpayers and file fake returns to steal refunds, according to Axios.
- Visa and Mastercard have reached a groundbreaking $5B settlement, reshaping the landscape of financial transactions as we know it.*
*This is a sponsored post. | | | |
| Upcoming Events | | MAR
27 | | Join DTC industry leaders and CEOs at the pivotal virtual roundtable event 'Shaping the Future of DTC in 2024.' | | | | | | | MAR
27 | | Buy-ology: How Today's Consumers Buy Online w/Mike Felix of Act-On | | | | | | | SEP
1 | | Join the $1M prize AI Challenge: Unleash your creativity and push AI Agents beyond today's limits. | | | | | | * This is a sponsored event | | | |
 | AI and technology writer | | Beth is a contributing editor and writer of Inside's AI and Tech newsletters. She has written for publications including USA Today, the Arizona Business Gazette, and The Arizona Republic, where she received recognition with a Pulitzer Prize nomination and a First Amendment Award for collaborative reporting on state pension cost increases. You can reach her at Beth.Duckett@yahoo.com. | | This newsletter was edited by Aaron Crutchfield | |
| |
| |
|
| |
|
| |
Comments
Post a Comment