TOKYO, Jul 29 (News On Japan) –
Hitachi has introduced a groundbreaking development in know-how to detect whether or not a doc was created by generative AI, using a world-first “triple electronic watermark” system geared toward stopping the unfold of pretend data.
As the web continues to be flooded with “fake news,” a detrimental byproduct of the widespread use of generative AI, Hitachi’s new know-how provides a major step ahead in combating this concern.
The system works by embedding particular phrases that perform as “electronic watermarks” inside the textual content generated by AI. The presence and frequency of those phrases assist decide if the doc was AI-generated.
A reporter demonstrated, “This text contains ‘watermarks’ that can identify if it was created by generative AI.”
The innovation addresses a crucial vulnerability: beforehand, a single digital watermark might be simply solid. Hitachi’s newest know-how makes use of a number of, layered digital watermarks, enhancing safety.
Koichi Nagatsuka, from Hitachi’s R&D Group, defined, “‘Electronic watermark’ technology essentially distinguishes between AI-generated and human-created text.”
In an period the place the proliferation of pretend data has turn out to be a major social concern, this know-how has garnered consideration for its potential to authenticate the creators of movies and paperwork.
Although digital watermarks have been successfully used to detect faux movies and pictures, making use of them to textual content has been difficult. Nagatsuka famous, “With text, changing even a single word can alter the entire meaning, making it difficult to implement.”
Hitachi plans to additional develop this know-how to stop the unfold of misinformation and fight fraud utilizing generative AI.
Source: TBS