The German regulator, KJM, has reduced the buffer for facial age estimation for the highest level of age assurance from 5 to 3 years. This means, instead of users needing to be estimated as over 23, users must now be estimated as over 21.
Originally approved by KJM in November 2021 for 16+ (erotic) content, in May 2022 KJM approved the use for facial age estimation for the highest level of assurance, covering 18+ adult content, with a buffer of 5 years. Now, KJM have reduced the buffer required by platforms from 5 to 3 years. This follows an increasing improvement in accuracy and performance of Yoti’s facial age estimation.
Germany has one of the most stringent regulatory environments in the world, in particular when it comes to accessing adult content online. Now, according to the KJM, under the German regulations, regulated businesses must set an age threshold with an agreed buffer of 3 years.
Over time, independent evaluations by NIST, the ACCS, as well as Yoti’s July 2025 white paper have shown significantly lower error rates and smaller Mean Absolute Errors (MAE), particularly around the ages of 17–20.
Yoti’s False Positive Rate* for ages 13-17 year olds being correctly estimated as over 21 is 0.6%. This means there is the potential for only around 6 minors our of 1,000 to be incorrectly estimated as over 21 with a 3 year buffer.
Yoti’s current MAE is 1.1 years for ages 13 – 17. Our True Positive Rate* (TPR) for 13-17 year olds being correctly estimated as under 21 is 99.4%. This gives regulators a very high level of confidence that it is very unlikely minors will be able to access adult content.
Yoti’s anti-spoofing liveness detection also ensures that under age website visitors will not be able to use the photo of an older person to access age restricted content.
To find out more about how our Age Estimation technology works, including performance across age, gender and skin tones, please read our most recent white paper.
* True Positive Rate – the probability that an actual positive will test positive, such as an 18 year old correctly estimated to be under 21.
* False Positive Rate – the probability that an actual negative will test positive, such as an 18 year old incorrectly estimated to be over 21.