English

ByteDance Seed

Type: AI company
Place: China — China / Singapore
Last checked: 2026-09-22
Next check due: 2027-03-21

This record is preserved as reference material. It is not a current contact destination and is excluded from Places to contact and recommendations.

Ways to reach them

No verified safety or disclosure channel found
Web form
Information only
Recipient
ByteDance Seed
Who may use it
People following the published contact instructions
What it accepts
Observed model behaviour; AI model flaws; AI-related harm
Information only. This link provides background or evidence; it is not a place to send a message.No contact action is available for this entry.
Evidence that this route accepts contact
Not found — general contact only. The Responsible AI page returned no readable body content when fetched (JS-rendered) and no contact details.
ByteDance operates a security response centre for its consumer products, but no AI-model-safety intake is published for Seed.
https://seed.bytedance.com/en/
Homepage
Homepage
Information only
Recipient
ByteDance Seed
Who may use it
People following the published contact instructions
What it accepts
Observed model behaviour; AI model flaws; AI-related harm
Information only. This link provides background or evidence; it is not a place to send a message.No contact action is available for this entry.
Evidence that this route accepts contact
ByteDance's AI safety responses in 2026 were driven by legal and commercial pressure rather than researcher reports: it pledged safeguards for Seedance 2.0 only after legal threats from Disney and Paramount over IP, and suspended a photo-to-voice feature over risk concerns. Both were reactive and externally forced. No evidence of a functioning researcher channel.
https://seed.bytedance.com/en/
No published frontier safety framework found
Published policy or framework
Information only
Recipient
ByteDance Seed
Who may use it
People following the published contact instructions
What it accepts
Observed model behaviour; AI model flaws; AI-related harm
Information only. This link provides background or evidence; it is not a place to send a message.No contact action is available for this entry.
Evidence that this route accepts contact
ByteDance's AI safety responses in 2026 were driven by legal and commercial pressure rather than researcher reports: it pledged safeguards for Seedance 2.0 only after legal threats from Disney and Paramount over IP, and suspended a photo-to-voice feature over risk concerns. Both were reactive and externally forced. No evidence of a functioning researcher channel.
https://seed.bytedance.com/en/direction/responsible_ai

What it does

A Responsible AI research direction within Seed; TechNode reported in August 2026 that ByteDance formed a new AI Data and Safety department. No named external-facing safety office.

Honest assessment

ByteDance's AI safety responses in 2026 were driven by legal and commercial pressure rather than researcher reports: it pledged safeguards for Seedance 2.0 only after legal threats from Disney and Paramount over IP, and suspended a photo-to-voice feature over risk concerns. Both were reactive and externally forced. No evidence of a functioning researcher channel.

Notes

Record as: no meaningful public safety channel for the Seed research arm. A concern about Seedance or Doubao output realistically has to go through the consumer product's in-app reporting, or through press and legal pressure — which is what has demonstrably moved them.

Sources

Something wrong here?