Home Tachnologies Amazon joins AI image creation fray with new model

Amazon joins AI image creation fray with new model

0
Amazon joins AI image creation fray with new model

Amazon is joining the AI image generation fray with the release of its Titan text-to-image AI model. Announced during the AWS re:Invent conference, Titan Image Generator can create “realistic, studio-quality images” and is supposed to have built-in guardrails against toxicity and bias. Titan isn’t a standalone app or website but a tool that developers can build on to make their own image generators powered by the model; to use it, developers will need access to Amazon Bedrock.

Swami Sivasubramanian, AWS vice president of database, analytics, and machine learning, previewed Titan Image Generator during his keynote, pointing to the model’s ability to not just create an image from a natural language prompt but also change out backgrounds. It’s aimed squarely at an enterprise audience, rather than the more consumer-oriented focus of well-known existing image generators like OpenAI’s DALL-E.

All images from Titan Image Generator will automatically include what Amazon said are invisible watermarks. Vasi Philomin, vice president for generative AI at AWS, told The Verge this is part of the voluntary commitments Amazon signed with the White House in July. 

“We wanted a way to mark an image as created with AI, and specifically made with Titan Image Generator that will not interfere with the visual, have no latency, and cannot be cropped or compressed out,” Philomin said in an interview. 

He added the watermark is not confined to the metadata of the file.

The catch, though, comes with how the invisible watermark is detected. Amazon created an API that people can connect to and then feed the image to check the image’s provenance. Philomin said this is by design; after all, Titan is not an end product but a model, so developers building with Titan Image Generator will choose how to provide that information to users. 

Sivasubramanian also announced the general availability of other Titan models: Titan Text Lite, a smaller model that is used for lighter text generation tasks like copywriting; and Text Express, which is for larger uses like conversational chat apps.

Amazon will also extend copyright indemnity to customers who use its Titan foundation models, including text-to-image. It will also provide legal cover to people who use any Amazon-created AI application, even if the app used a different foundation model found in its Bedrock AI model repository, like Meta’s Llama 2 or Anthropic’s Claude 2. These apps include AWS HealthScribe, CodeWhisperer, Amazon Personalize, Amazon Lex, and Amazon Connect Contact Lens.