Image by Matthieu Rochette, from Unsplash

Researchers Test How To Store Data On A Bird

• Written by Kiara Fabbri Former Tech News Writer
• Fact-Checked by Sarah Frazier Former Content Manager

In a bizarre new experiment, a researcher discovered that birds can store and reproduce complex data, including pictures, using just sound.

In a rush? Here are the quick facts:

• European starling mimics ultrasonic sounds to encode complex image data.
• Researchers used a spectrogram synthesizer to convert images into sounds.
• The starling replicated simplified sound patterns in the same frequency range.

The YouTuber Ben Jordan, who led this experiment, did so by observing a European starling nicknamed “The Mouth,” known for its remarkable ability to mimic sounds with high ultrasonic accuracy.

Ben explains in his video that it achieved this by converting a simple bird drawing into a series of high-frequency tones using a special tool called a spectrogram synthesizer.

“I drew a photo of a bird in a spectral synthesizer and put the sound on my phone to play for the Starling on the off chance that he’d like the sound enough to add to his vocabulary, which would effectively make himself capable of storing and transmitting image data.” Ben explained.

The researchers aimed to determine if the bird could learn and reproduce the sound, essentially encoding the image as vocal sounds.

The results were positive. When they listened to ultrasonic recordings, they found that the starling had copied a simplified version of the sound pattern.

“This little bird successfully learned and emulated the sound in the exact same frequency range that he heard it, effectively transferring about 176 kilobytes of uncompressed information.” Ben said. Although not perfect, this showed that birds could store and share data through their calls.

Ben explains key to the discovery was special audio gear that can capture sounds too high for humans to hear, including 192kHz ultrasonic microphones, time-stretching software to slow down the sounds without losing quality, and spectrogram analysis to visualize the bird’s singing.

Ben also noted, “Hypothetically, if this were an audible file transfer protocol that used a 10:1 data compression ratio, that’s nearly 2 megabytes of information per second”

For those interested in trying similar experiments, Ben recommends affordable tools like the AudioMoth ultrasonic recorder, Raspberry Pi with BirdNET-Pi software, and the free Merlin app by Cornell Lab.

Ben summed it up: “The fact that you could set up a speaker in your yard and conceivably store any amount of data in songbirds is crazy.”

Image by franco alva, from Unsplash

Meta Sued For Allegedly Using Porn To Train AI

• Written by Kiara Fabbri Former Tech News Writer
• Fact-Checked by Sarah Frazier Former Content Manager

Adult film producers Strike 3 Holdings and Counterlife Media, have launched a major lawsuit against Meta, accusing the tech giant to have illegally downloaded and distributed almost 2,400 copyrighted adult films in order to train its AI models.

In a rush? Here are the quick facts:

• Strike 3 Holdings claims Meta used BitTorrent to download 170+ copyrighted videos.
• Evidence links downloads to Meta IP addresses and an employee’s home connection.
• Plaintiffs seek $359 million and deletion of infringing AI data.

The California federal court lawsuit – first retrieved by ArsTechnica – states that Meta used the BitTorrent peer-to-peer file-sharing system to download pirated content, since at least 2018. The adult sites claim that Meta downloaded the videos while simultaneously sharing them over weeks or months.

This tactic, they allege, helped Meta download other files faster due to BitTorrent’s “tit-for-tat” system, which rewards users who share more popular content.

“Meta specifically targeted Plaintiffs’ content for distribution in order to accelerate its downloads of vast amounts of other content,” the lawsuit says.

Strike 3 argues that Meta’s actions gave it an unfair competitive advantage, because its AI models learned from high-quality porn content, which could result in the creation of similar adult content.

“By training so specifically on Plaintiffs’ Works, Meta’s AI Movie Gen may very well soon produce full length films with Plaintiffs’ identical style and quality,” they said.

The lawsuit also raises concerns that Meta’s piracy may have let minors access adult films without age checks, in violation of new state laws. “Plaintiffs cannot compete against Meta when it ignores federal and state laws and offers Plaintiffs’ works for free,” Strike 3 alleged.

The evidence shows more than 100,000 unauthorized distribution instances linked to Meta’s IP addresses through “stealth” networks, and even from a Meta employee’s personal home network. Meta states “We’re reviewing the complaint, but don’t believe Strike’s claims are accurate,” as reported by ArsTechnica.

TorrentFreak notes that the court could impose penalties of up to $359 million on Meta while requiring the company to remove pirated content from its AI training data.