Under the envisioned framework, both candidate and issue ads would be required to include an on-air and filed disclosure that AI-generated content was used.
© 2024 TechCrunch. All rights reserved. For personal use only.
Gemini, the company’s family of generative AI models, will enhance the smart TV operating system so it can generate descriptions for movies and TV shows.
© 2024 TechCrunch. All rights reserved. For personal use only.
Veo can capture different visual and cinematic styles, including shots of landscapes and timelapses, and make edits and adjustments to already-generated footage.
© 2024 TechCrunch. All rights reserved. For personal use only.
The official launch comes almost a year after YouTube began experimenting with AI-generated quizzes on its mobile app.
© 2024 TechCrunch. All rights reserved. For personal use only.
The Israeli startup has raised $5.5M for its platform that uses “statistical AI” to generate synthetic data that it says is as good as the real thing.
© 2024 TechCrunch. All rights reserved. For personal use only.
TikTok is starting to automatically label AI-generated content that was made on other platforms, the company announced on Thursday. With this change, if a creator posts content on TikTok that was created with a service like OpenAI’s DALL·E 3, it will automatically have an “AI-generated” label attached to it to notify viewers that it was […]
© 2024 TechCrunch. All rights reserved. For personal use only.
OpenAI’s video generation tool Sora took the AI community by surprise in February with fluid, realistic video that seems miles ahead of competitors. But the carefully stage-managed debut left out a lot of details — details that have been filled in by a filmmaker given early access to create a short using Sora.
Shy Kids is a digital production team based in Toronto that was picked by OpenAI as one of a few to produce short films essentially for OpenAI promotional purposes, though they were given considerable creative freedom in creating “air head.” In an interview with visual effects news outlet fxguide, post-production artist Patrick Cederberg described “actually using Sora” as part of his work.
Perhaps the most important takeaway for most is simply this: While OpenAI’s post highlighting the shorts lets the reader assume they more or less emerged fully formed from Sora, the reality is that these were professional productions, complete with robust storyboarding, editing, color correction, and post work like rotoscoping and VFX. Just as Apple says “shot on iPhone” but doesn’t show the studio setup, professional lighting, and color work after the fact, the Sora post only talks about what it lets people do, not how they actually did it.
Cederberg’s interview is interesting and quite non-technical, so if you’re interested at all, head over to fxguide and read it. But here are some interesting nuggets about using Sora that tell us that, as impressive as it is, the model is perhaps less of a giant leap forward than we thought.
Control is still the thing that is the most desirable and also the most elusive at this point. … The closest we could get was just being hyper-descriptive in our prompts. Explaining wardrobe for characters, as well as the type of balloon, was our way around consistency because shot to shot / generation to generation, there isn’t the feature set in place yet for full control over consistency.
In other words, matters that are simple in traditional filmmaking, like choosing the color of a character’s clothing, take elaborate workarounds and checks in a generative system, because each shot is created independent of the others. That could obviously change, but it is certainly much more laborious at the moment.
Sora outputs had to be watched for unwanted elements as well: Cederberg described how the model would routinely generate a face on the balloon that the main character has for a head, or a string hanging down the front. These had to be removed in post, another time-consuming process, if they couldn’t get the prompt to exclude them.
Precise timing and movements of characters or the camera aren’t really possible: “There’s a little bit of temporal control about where these different actions happen in the actual generation, but it’s not precise … it’s kind of a shot in the dark,” said Cederberg.
For example, timing a gesture like a wave is a very approximate, suggestion-driven process, unlike manual animations. And a shot like a pan upward on the character’s body may or may not reflect what the filmmaker wants — so the team in this case rendered a shot composed in portrait orientation and did a crop pan in post. The generated clips were also often in slow motion for no particular reason.
Example of a shot as it came out of Sora and how it ended up in the short. Image Credits: Shy Kids
In fact, using the everyday language of filmmaking, like “panning right” or “tracking shot” were inconsistent in general, Cederberg said, which the team found pretty surprising.
“The researchers, before they approached artists to play with the tool, hadn’t really been thinking like filmmakers,” he said.
As a result, the team did hundreds of generations, each 10 to 20 seconds, and ended up using only a handful. Cederberg estimated the ratio at 300:1 — but of course we would probably all be surprised at the ratio on an ordinary shoot.
The team actually did a little behind-the-scenes video explaining some of the issues they ran into, if you’re curious. Like a lot of AI-adjacent content, the comments are pretty critical of the whole endeavor — though not quite as vituperative as the AI-assisted ad we saw pilloried recently.
The last interesting wrinkle pertains to copyright: If you ask Sora to give you a “Star Wars” clip, it will refuse. And if you try to get around it with “robed man with a laser sword on a retro-futuristic spaceship,” it will also refuse, as by some mechanism it recognizes what you’re trying to do. It also refused to do an “Aronofsky type shot” or a “Hitchcock zoom.”
On one hand, it makes perfect sense. But it does prompt the question: If Sora knows what these are, does that mean the model was trained on that content, the better to recognize that it is infringing? OpenAI, which keeps its training data cards close to the vest — to the point of absurdity, as with CTO Mira Murati’s interview with Joanna Stern — will almost certainly never tell us.
As for Sora and its use in filmmaking, it’s clearly a powerful and useful tool in its place, but its place is not “creating films out of whole cloth.” Yet. As another villain once famously said, “that comes later.”
The Oversight Board, Meta’s semi-independent policy council, it turning its attention to how the company’s social platforms are handling explicit, AI-generated images. Tuesday, it announced investigations into two separate cases over how Instagram in India and Facebook in the U.S. handled AI-generated images of public figures after Meta’s systems fell short on detecting and responding to the explicit content.
In both cases, the sites have now taken down the media. The board is not naming the individuals targeted by the AI images “to avoid gender-based harassment,” according to an e-mail Meta sent to TechCrunch.
The board takes up cases about Meta’s moderation decisions. Users have to appeal to Meta first about a moderation move before approaching the Oversight Board. The board is due to publish its full findings and conclusions in the future.
Describing the first case, the board said that a user reported an AI-generated nude of a public figure from India on Instagram as pornography. The image was posted by an account that exclusively posts images of Indian women created by AI, and the majority of users who react to these images are based in India.
Meta failed to take down the image after the first report, and the ticket for the report was closed automatically after 48 hours after the company didn’t review the report further. When the original complainant appealed the decision, the report was again closed automatically without any oversight from Meta. In other words, after two reports, the explicit AI-generated image remained on Instagram.
The user then finally appealed to the board. The company only acted at that point to remove the objectionable content and removed the image for breaching its community standards on bullying and harassment.
The second case relates to Facebook, where a user posted an explicit, AI-generated image that resembled a U.S. public figure in a Group focusing on AI creations. In this case, the social network took down the image as it was posted by another user earlier, and Meta had added it to a Media Matching Service Bank under “derogatory sexualized photoshop or drawings” category.
When TechCrunch asked about why the board selected a case where the company successfully took down an explicit AI-generated image, the board said it selects cases “that are emblematic of broader issues across Meta’s platforms.” It added that these cases help the advisory board to look at the global effectiveness of Meta’s policy and processes for various topics.
“We know that Meta is quicker and more effective at moderating content in some markets and languages than others. By taking one case from the US and one from India, we want to look at whether Meta is protecting all women globally in a fair way,” Oversight Board Co-Chair Helle Thorning-Schmidt said in a statement.
“The Board believes it’s important to explore whether Meta’s policies and enforcement practices are effective at addressing this problem.”
Some — not all — generative AI tools in recent years have expanded to allow users to generate porn. As TechCrunch reported previously, groups like Unstable Diffusion are trying to monetize AI porn with murky ethical lines and bias in data.
In regions like India, deepfakes have also become an issue of concern. Last year, a report from the BBC noted that the number of deepfaked videos of Indian actresses has soared in recent times. Data suggests that women are more commonly subjects for deepfaked videos.
Earlier this year, Deputy IT Minister Rajeev Chandrasekhar expressed dissatisfaction with tech companies’ approach to countering deepfakes.
“If a platform thinks that they can get away without taking down deepfake videos, or merely maintain a casual approach to it, we have the power to protect our citizens by blocking such platforms,” Chandrasekhar said in a press conference at that time.
While India has mulled bringing specific deepfake-related rules into the law, nothing is set in stone yet.
While the country there are provisions for reporting online gender-based violence under law, experts note that the process could be tedious, and there is often little support. In a study published last year, the Indian advocacy group IT for Change noted that courts in India need to have robust processes to address online gender-based violence and not trivialize these cases.
There are currently only a few laws globally that address the production and distribution of porn generated using AI tools. A handful of U.S. states have laws against deepfakes. The UK introduced a law this week to criminalize the creation of sexually explicit AI-powered imagery.
In response to the Oversight Board’s cases, Meta said it took down both pieces of content. However, the social media company didn’t address the fact that it failed to remove content on Instagram after initial reports by users or for how long the content was up on the platform.
Meta said that it uses a mix of artificial intelligence and human review to detect sexually suggestive content. The social media giant said that it doesn’t recommend this kind of content in places like Instagram Explore or Reels recommendations.
The Oversight Board has sought public comments — with a deadline of April 30 — on the matter that addresses harms by deep fake porn, contextual information about the proliferation of such content in regions like the U.S. and India, and possible pitfalls of Meta’s approach in detecting AI-generated explicit imagery.
The board will investigate the cases and public comments and post the decision on the site in a few weeks.
These cases indicate that large platforms are still grappling with older moderation processes while AI-powered tools have enabled users to create and distribute different types of content quickly and easily. Companies like Meta are experimenting with tools that use AI for content generation, with some efforts to detect such imagery. However, perpetrators are constantly finding ways to escape these detection systems and post problematic content on social platforms.
Founded in 2017, Atoms Lanka Solutions is a provider of IT consulting and software development services. Throughout the company’s history, we have worked with over 160 customers from 5+ countries..
