Facebook saw a 134% jump in the number of hate posts it removed in Q2 2020 compared to Q1, according to its latest Community Standards Enforcement Report. The company attributed the increase to better AI detection systems and improved automation technology for its Spanish, Arabic, and Indonesian posts.

More:

• Facebook has turned more to AI reviewers to flag and remove harmful, dangerous, and false content during the pandemic.
• The company removed 9.6 million hateful posts in Q1 and 22.5 million in Q2. Its AI systems detected 84% of known hate speech in Q1, compared to 45% in Q2.
• On Instagram, the company removed 3.3m hate posts in Q2 versus 808,900 in Q1.
• Facebook's proactive rate — the percentage of violating content that AI and human reviewers find before users do - has steadily increased since 2017 (see graph above).

Related:

• A company-commissioned audit made public in July said Facebook should improve its AI tools for identifying hate speech and other dangerous content.
• Facebook continues to fall short in its efforts to identify algorithmic bias, according to the report, which focused on civil-rights issues on the platform.
• Facebook is not attuned enough to the way its algorithms inadvertently feed "extreme and polarizing content,” it added.
• In June, Facebook apologized after its AI accidentally blocked posts featuring an 1890s image of Aboriginal men in chains, citing it as inappropriate nudity.

A new class-action lawsuit accuses Facebook of illegally gathering biometric data through Instagram's photo-tagging tool. Facebook faces up to $500b in fines from the lawsuit, which was filed in California this week.

More:

• The lawsuit claims Facebook-owned Instagram illegally collects, stores, and profits from the biometric data of its 100m-plus users. It alleges that Instagram illegally uses a facial recognition tool to scan the faces of people pictured in others' posts, even if they don't use the platform.
• Illinois resident Kelly Whalen filed the lawsuit on Monday in San Mateo Superior Court. She is asking that Facebook pay $1,000 to $5,000 for each violation of Illinois' biometric privacy law, which bars the scanning of biometric data of individuals without written consent.
• A Facebook spokeswoman denied that Instagram uses facial recognition technology.
• From Twitter: @Scotch_man wrote: "How could you have done this, company who has been known to do this repeatedly and unabashedly since inception- how could you?"

Related:

• Last month, Facebook agreed to settle another facial recognition lawsuit for $650m, one of the largest payouts ever related to data privacy.
• That lawsuit argued Facebook violated the Illinois privacy law on biometric data privacy through the use of its “tag suggestions” feature.
• The settlement is $100m more than Facebook's initial offer of $550m, which a U.S. district judge rejected in June.

Report: AI Skills in the U.K.

U.K. companies aren't doing enough to prepare employees for increased use of AI, according to a newly released report.

The findings:

• 15% of U.K. companies are classified as advanced "AI pros," compared to 23% of global companies.
• 52% of U.K. employees are using AI to work faster and smarter compared to 69% of employees globally.
• 35% of U.K. business leaders foresee an AI skills gap in the next two years; 28% say it's already here.
• 17% of U.K. employees are being re-skilled for AI compared to 38% globally...

AI drug discovery startup Atomwise raised $123m to scale its team and machine learning platform. Atomwise uses AI and machine learning algorithms to analyze data on the creation of new drug molecules. Its neural network, AtomNet, can analyze more than 100 million compounds per day.

More:

• Companies that use the drug discovery platform include Eli Lilly and Company, Jiangsu Hansoh Pharmaceutical Group Co. Ltd., Bayer AG, and Bridge Biotherapeutics Inc. Atomwise's joint-venture investments now total $5.5b.
• The Series B financing was led by B Capital Group and Sanabil Investments, with participation from DCVC, BV, Tencent, Y Combinator, Dolby Ventures, and AME Cloud Ventures.
• San Francisco-based Atomwise raised nearly $175m since its 2012 inception.
• The bigger picture: The AI drug discovery market is expected to reach $40b by 2027.

A GitHub user has posted the open-source code for AnimeGAN, which turns photos into anime-style background art. The AI system comes from Wuhan University and Hubei University of Technology researchers.

More:

• AnimeGAN uses machine learning through neural style transfer and generative adversarial networks.
• The researchers used films from Hayao Miyazaki, Makoto Shinkai, and Satoshi Kon to train the AI.
• Their paper, "AnimeGAN: A Novel Lightweight GAN for Photo Animation,” was submitted to the International Symposium on Intelligence Computation and Applications.

Self-driving truck startup Rivian asked a judge to dismiss a lawsuit filed by Tesla over allegations the startup stole trade secrets by poaching former Tesla employees. Rivian argues that Tesla’s claims are unsubstantiated. Lawyers for the startup said Tesla’s lawsuit was intended to damage Rivian’s reputation and threaten Tesla employees who decide to leave the automaker.

More:

• Rivian said it has not found any stolen trade secrets within any of its systems, adding it has rigorous procedures to ensure that it does not obtain outside confidential information.
• Kristin Korosec of TechCrunch noted that Rivian did not dispute Tesla’s complaint, but instead accused Tesla of wrongdoing.
• The lawsuit, filed by Tesla in July, claimed that Rivian hired 178 former Tesla employees in an effort to steal its secrets. Most of its former employees are still held to confidentiality agreements.
• Rivian plans to launch a series of electric vehicle models in 2021 that will have Level 3 autonomy when vehicles can drive themselves in favorable road conditions.

A version of this story first appeared in Inside Electric Vehicles.

QUICK HITS

• Palantir is reportedly planning to go public via a direct listing in September.
• Deloitte Consulting won a $106m contract to build the AI development platform for the Pentagon’s artificial intelligence hub.
• Google developed an AI system that can detect orca sounds to protect the whales from ships.
• IBM released three online tools to teach young people about the future of AI.
• The fittest athletes in the world swear by CBD. See what they're using here.*

*This is a sponsored post.

Tweet of the Day: Nvidia's Bojan Tunguz shared this comic about machine learning arguments.