Amid the craziest news cycle in recent memory, AI-generated deepfakes have yet to become the huge truth catastrophe that experts warned about.

Why it matters: Media outlets and tech platforms have gotten better at spotting and debunking AI misinformation quickly.

• Most falsehoods that go viral during breaking news are still created not through doctored media, but through manipulated context, such as the wrong caption describing when a photo was taken.

Case in point: In the wake of the Trump assassination attempt, many mainstream news outlets and wire services, including Reuters, The AP, Politico, BBC and CNN, quickly posted fact checks debunking the validity of a doctored image that appeared to show Secret Service agents smiling while lifting Trump to his feet on stage after the shooting.

• Several other fact-checking sites, including FactCheck.org, Verify and PolitiFact were also able to quickly debunk the photo using very standard fact-checking methods, such as reverse image search results.

Zoom in: The Biden campaign felt comfortable enough with people's trust in social media that he posted the announcement of his withdrawal from the presidential race on X, Facebook, Instagram and Threads before alerting traditional media.

Yes, but: Doctored photos and video can still break through if tech platforms aren't quick to police them.

• The fake image of the smiling secret service agents still exists on X, formerly Twitter, but now with a Community Note attached.
• While these Community Notes can be helpful, they're sometimes not added quickly or broadly enough to stop misinformation from spreading.
• Musk has made X a breeding ground for a lot of misinformation that gets picked up and spread on smaller platforms, where it goes viral.
• A highly implausible deepfake PBS News video of Biden cursing is also still circulating on X, but PBS was quick to point out that it isn't real.

Broader conspiracies that have little to do with doctored photos or videos have proven even harder to stop.

• In the wake of the Trump assassination, conspiracies about who the gunman was and the motivation behind the shooting gained plenty of steam online, but most of those falsehoods weren't predicated on doctored photos.

What to watch: While AI-generated deepfakes have become easier for news outlets and fact-checkers to debunk in real time, customized responses to AI chatbots about breaking news are much harder to police.

A new study showcases significant advances in weather and climate modeling, achieved by taking AI technologies and fusing them with some components of standard, physics-based models.

Why it matters: Its creators say the new model, dubbed "NeuralGCM," has proven to be more accurate than other purely machine learning-based models for one- to 10-day weather forecasts, along with the top extended-range models in use today.

• It also has proven uniquely skillful in projecting climate conditions over decades, according to the findings, published Monday in the journal Nature.

Zoom in: The findings demonstrate how quickly the field of AI-based weather and climate forecasting is advancing.

• AI models have tremendous compute power and timeliness advantages over traditional computer models.
• The new model, for example, is open source, and is designed to be run relatively quickly on a laptop, study co-author Stephen Hoyer of Google Research told Axios.
• By contrast, traditional weather forecasting models take hours to run on the world's most powerful supercomputers in order to work through tens of thousands of lines of code describing the physical laws of how the atmosphere and oceans work.

How it works: The new model, from scientists at Google Research, Google DeepMind, MIT, Harvard University and the European Centre for Medium-Range Weather Forecasts, uses machine learning and a neural network.

• This loosely models neurons in the brain to train off of decades of past weather data.
• It also uses physics equations describing large-scale weather patterns, essentially combining a global circulation model with its physics-intensive approach with AI-driven tasks.

Aaron Hill, an assistant meteorology professor at the University of Oklahoma, told Axios that one of the biggest "novelties" of the new model is how it keeps some of the large-scale physics and replaces some parts of the modeling with AI.

• AI forecasting models made by Nvidia, Microsoft and other companies do away with the physics altogether, he noted.

Between the lines: Hill, who wasn't involved in the new study, said AI and machine learning techniques are rapidly being adopted in the weather and climate research communities.

• But they haven't yet made the leap into operations, meaning day-to-day, public-facing forecasting at agencies like the National Oceanic and Atmospheric Administration or its international counterparts.

• Five Senate Democrats want answers about OpenAI's safety efforts after several employees warned earlier this month that OpenAI rushed through safety testing. (Washington Post)
• Senate Majority Leader Chuck Schumer is bringing two kids online safety bills up for a vote after working with LGBTQ+ rights advocates to refine them for over a year. (Axios Pro)
• Google will no longer phase out third-party tracking cookies. (Axios)
• Wiz said it is walking away from a Google deal reportedly valued at $23 billion, and will instead pursue an IPO. (CNBC)
• With 78 million daily active users exchanging over 50,000 chat messages every second, Roblox struggles to keep kids safe from online predators on its platform. (Bloomberg)

Here's me at the Élysée Palais, the French equivalent of the White House, where Macron held a reception Monday for foreign journalists covering the Olympics.