During a keynote at its re:Invent conference this morning, Amazon debuted Titan Image Generator, which is now available in preview for AWS customers. Titan Image Generator, a part of Amazon’s Titan family of generative AI models, can create new images given a text description or customize existing images.
“[You] can use the model to easily swap out an existing background to a background of a rainforest [for example],” Swami Sivasubramanian, the VP for data and machine learning services at AWS, said onstage. “[And you] can use the model to seamlessly swap out backgrounds to generate lifestyle images, all while retaining the main subject of the image and to create a few more options.”
Amazon says that Titan Image Generator was trained on a “diverse set of data sets” across a “broad range of domains” (and can be optionally fine-tuned on custom data sets) and includes built-in mitigations for toxicity and bias. The company declined to say exactly where those data sets came from, however — and whether it obtained permission from or is compensating all the creators of the images that it used to train Titan Image Generator.
Sivasubramanian did claim onstage, however, that Amazon will protect customers accused of violating copyright with images created using Titan Image Generator — in keeping with its AI indemnification policy.
“We’re carefully choosing how we train our models and the data we use to do so,” he said. “With these investments, our customers will have the flexibility to select the best models for their requirements even as their needs grow and change.”
Images created with Titan Image Generator will also come with a “tamper-resistant” invisible watermark by default — an attempt to mitigate the spread of AI-generated misinformation, Sivasubramanian says. (Presumably, customers can switch watermarking off though.)