May 25, 2018 marks an important day in European history. It was the day the General Data Protection Regulation (GDPR) came into effect. “From that point on, it became more difficult for companies to share data. Projects went on hold," remembers Marian Gläser, co-founder and managing director of the Berlin start-up brighter AI. "This presented a problem for the entire industry." But for every problem there is a solution. The Berlin-born visionary and his team spotted an opportunity to help companies overcome the issue: brighter AI's innovative Deep Natural Anonymisation solution recognises personal features such as faces or license plates recorded by cameras.
"It generates synthetic image data in real time, with the help of generative artificial intelligence, which is superimposed on the original data," explained Gläser in an interview . "The synthetic overlays look real, allowing machine learning algorithms to be applied to them, but cannot be traced back to a specific individual and therefore do not provide personally identifiable information about the people depicted." In the processed images, although a person with apparently normal facial features can still be recognised, the actual person is anonymised and no longer identifiable. This ensures video recordings that comply with the strict data protection guidelines stipulated by the GDPR in Europe, but also the California Consumer Privacy Act (CCPA), and protect the identities of people in public spaces.
The innovative solution allows companies to retain data quality for digital analytics and machine learning, which distinguishes it from other redaction techniques that attempt to ensure privacy through pixelation or black bars. Thanks to brighter AI's anonymisation technology, "data protection should no longer be an issue", adds the Berliner. "We take care of data protection and offer a solution that combines privacy and technical innovation, allowing companies to concentrate on their goals."
Strategic change of course: from night vision to data protection
Brighter AI owes its success more to a smart change of course than to long-term planning. The Berlin start-up, which was founded in 2017 as a spin-off from the incubator of the automotive supplier and lighting specialist HELLA, originally had a different vision: The team led by Marian Gläser and co-founder Patrick Kern developed neural networks that generated realistic daylight images from night recordings of infrared cameras in autonomous vehicles. That is until 25 May 2018.
Since then, it has been legally prohibited to collect the necessary data on the street and share it with cooperation partners. "The fact that we decided to change course one year after the spin-off is the most exciting thing that could have happened to us," says the IT expert Gläser, reflecting on the strategic change that led to the young company being named Europe’s Hottest AI Start-Up. “We have always strived to develop industrial solutions to empower people. That has stayed the same. Only our direction has changed, which has created new opportunities."
The brighter AI team of 20 quickly noticed that the challenges presented by the newly introduced data protection guidelines affected not just the automotive sector. "As soon as we pitched our solution, we received enquiries from retail, medicine and smart city," says Gläser. The Berlin start-up's anonymisation technology is now used in areas such as hazard detection, intelligent store concept development and smart traffic planning. Its customers include Intel, Valeo, Facebook and Deutsche Bahn.
Deutsche Bahn: cameras against coronavirus
"According to Deutsche Bahn's mindbox programme, we were the first start-up that enabled it to work directly with video data, thanks to our data protection solution," says Gläser, proud of another historic brighter AI achievement. Despite the approximately 40,000 cameras that are used in trains and on platforms, the company has so far only been able to use the recorded data for security issues or to a very limited extent. However, this information is important in order to intelligently manage the flow of people and thus optimise the infrastructure and operational processes and improve the overall customer and travel experience.
"Another aspect was to understand how people are distributed on trains in order to better distribute the load in individual wagons," adds Gläser, who is a spokesman for AI in the Federal Association of German Start-Ups. It is an issue that has skyrocketed in relevance due to the current social distancing guidelines to curb the spread of Covid-19. Since autumn 2020, the company has been using brighter AI technology as a solution: After a successful pilot project, the technology has recently been installed in selected trains and the solution is expected to be rolled out nationwide in the near future.
Europe: a role model for social change
Deutsche Bahn is not the only company that relies on brighter AI technology. The award-winning start-up, whose honours include the Deep Tech Award 2020 and The Spark - The German Digital Award from Handelsblatt & McKinsey, also has well-known customers France, Sweden or Austria. "Thanks to the GDPR, Europe has great potential to become the market leader in privacy tech," says Gläser, who sees the regulations in a positive light. “Companies like us that work on solutions have a head start on the market and can give our customers a competitive edge.” This advantage is appreciated not only by European companies.
American and Japanese corporations have also shown great interest in anonymisation technology from the German capital. In addition, there are a growing number of enquiries from China and from European companies that collect data in the People's Republic and export it to Europe in compliance with data protection regulations. In order to meet this increasing demand, brighter AI extended its range about six months ago. The software no longer needs to be installed in the customer's in-house data centre - as it was previously the case - but is now available on a centralised cloud platform. "Whether it's via a company's data centre or with us - the experience and quality are the same,” emphasises Gläser, pleased to provide a service that caters not only to large enterprises. "Even YouTubers are now starting to anonymise data," which shows that new customer groups can be reached with B2B2C models. This alone results in almost 1,000 hours of video material in the cloud, which the AI processes. "In the last year and a half, we have been able to almost double the data that we anonymise every quarter," says the brighter AI co-founder.
Berlin: an international tech hub
The success story of brighter AI, whose investors include eCAPITAL Entrepreneurial Partners and Giesecke + Devrient Ventures, is an example of the benefits of establishing a tech start-up in German capital. 28 per cent of the German companies in the AI sector have settled in Berlin-Brandenburg, for start-ups alone the figure rises to about 50 per cent. In a recent study conducted by the Berlin Technology Foundation, 223 AI companies were identified that employ around 5,000 people and generate sales of almost € 500 million. By 2025, annual sales of Berlin-based AI companies are expected to grow to over € 2 billion.
Berlin is also home to many important AI research institutions. A field that brighter AI knows well: "We work with six different universities," says Gläser. "We are currently working on our second project with the Fraunhofer Institute." As a Berliner at heart, the founder raves about the city's good ecosystem of start-ups and AI companies as well as the creative knowledge exchange. “Founding a company in Berlin and being in this environment offers us many advantages internationally,” says Gläser. “For the past four or five years, Berlin has been recognised as a major hub of innovation. Large corporations have established their innovation units here. Even Tesla is coming to Berlin-Brandenburg region." Domestic start-ups and companies, in turn, are growing and scaling up from the German capital into the world. Brighter AI is also planning to open a second site in the USA in the near future.
Goal: "To protect every identity in the public space"
The step across the pond should bring brighter AI closer to its goal of "protecting every identity in the public space". This is not just an empty promise, as a small but important project for Marian Gläser shows: On its pro bono platform protect.photo, private individuals can have up to four individual images anonymised free of charge.
“This allows people to upload pictures of demonstrations such as Black Lives Matter without revealing their identity. You can fight for something without being followed," says the IT expert. "When pixelation is used, images are shared less because they are less emotional. With our technology, the emotion is preserved, while the points of view of people are protected." However, this example of brighter AI aimed at individuals is an exception. “We will remain a B2B company,” emphasises Gläser. "With large companies collecting images in public and taking three or four recordings of each person, we are very, very close to delivering on our promise to protect every identity."