The Focused Forest - Vol.5

The Focused Forest - Vol.5

Number 5 in this weekly AI roundup is here.

European Union introduces the AI Act.

The EU has introduced the AI Act, the worlds first comprehensive regulation on AI which essentially aims to ensure consistent standards for AI systems across member states by focusing on safeguarding health, safety and rights of EU citizens and address risks associated with AI.

The AI Act classifies AI based on the risk they pose to users, with different risk levels subject to varying degrees of regulation. It also emphasizes the importance of human oversight to prevent harmful outcomes and requires transparency, traceability, and non-discrimination in the use of AI systems.


The AI Act encompasses several key provisions:

  • Risk - based approach: AI systems will be classified into four different risk classes, each carrying different levels of regulation. Most notably the following are considered a threat to people and will be banned: "cognitive behavioural manipulation of people or specific vulnerable groups, social scoring, real time and remote biometric identification systems.
  • Safety & Fundamental Rights: Ensure the safety of AI systems on the EU market and respect fundamental rights and EU values.
  • Transparency & Traceability: Aims to prevent harmful outcomes and provide legal certainty for investments and innovations in AI.
  • Human Oversight: The AI Act emphasizes the importance of human oversight in the use of AI systems to prevent harmful outcomes
  • Enforcement and Governance: it is expected that the AI act will be enforced through national market surveillance in each member state with the establishment of an AI office for the EU.

AI regulation has been one of the hottest topics in 2023 with the US, UK and EU in heavy dialogue with industry executives and lawmakers in an effort to find common ground and not stifle innovation and competition.

France and Germany have warned against excessive regulation as they look to protect the interests of their world leading AI startups.

“We can decide to regulate much faster and much stronger than our major competitors. But we will regulate things that we will no longer produce or invent. This is never a good idea... When I look at France, it is probably the first country in terms of artificial intelligence in continental Europe. We are neck and neck with the British. They will not have this regulation on foundational models."

The EU will regulate foundational models by requiring developers to provide documentation on model training methods and the data used. Failure to comply will mean fines of €35 million or 7 per cent of global revenue.

France Digitale are an independent organisation which represents European startups and investors and has warned against over regulation and voiced particular concern for products within the high-risk category which will require a CE mark, a long and expensive process which could hurt startups. Although they welcomed the flexibility in the policy which allows businesses to petition against the status.

"We called for not regulating the technology as such, but regulating the uses of the technology. The solution adopted by Europe today amounts to regulating mathematics, which doesn't make much sense," the group said."


Mistral AI joins the AI ?? Club

French generative AI startup Mistral have joined the prestigious AI unicorn club, they are now valued at $2 billion after a $415 million raise led by Andreessen Horowitz and Lightspeed Ventures.

Mistral focus on open-source technology for generative AI tools and chatbots and aim to rival US giants OpenAI, Google and Meta.

While most startups release their products with extravagant announcements, blog posts and press releases, Mistral released their latest LLM out of the blue via a Torrent link on X.

https://mistral.ai/news/mixtral-of-experts/

The model outperforms some of its U.S. competitors, including Meta’s Llama 2 family and OpenAI’s GPT-3.5.


Pro-China network used YouTube to Malign US.

The New York Times reported last week that a pro-China YouTube network used artificial intelligence to Malign the US and promote Chinese interests in technology, the Middle East and Russia.


More than 4500 videos published across 30 YouTube channels spread pro- Chinese and anti-US narratives according to a report published by the Australian Strategic Policy Institute, a security focused think tank.

https://www.aspi.org.au/index.php/report/shadow-play

The Campaign has been dubbed the "Shadow Play" and has attracted around 120 million views and 730,000 subscribers and began publishing content in mid-2022.


Atomic AI creates LLM using chemical mapping data for RNA drug discovery

San Francisco biotech startup Atomic AI have developed a world fist LLM based on chemical mapping data to optimize RNA targeting drugs. Published in bioRxiv their ATOM-1 platform is a foundational model used to accurately predict structure and function of RNA.

ATOM-1 was developed on large scale clinical mapping data collected in house using custom wet-lab assays across data on millions of RNA sequences and a billion nucleotide-level measurements.

“By building large datasets based on RNA nucleotide modifications and next-generation sequencing, the team at Atomic AI has created a first-of-its-kind RNA foundation model,” said Stephan Eismann, Ph.D., Founding Scientist and Machine Learning Lead at Atomic AI.

Link to the paper below.


Venture

MedTech, biotechnology and defence all saw big deals, with only a couple of weeks left of the year.


Lazada - Singapore based ecommerce site raised $630 million via 阿里巴巴集团 . This is their third round of 2023 bringing this years total to over 1.8 billion.

Tome Biosciences raised a massive $213 million to emerge from stealth with a combined series A and B round led by Andreessen Horowitz and Arch Venture Partners. Their molecular surgery tech is designed to allow the insertion of varying sizes genetic material without damaging the DNA focused on liver and autoimmune diseases.

True Anomaly raised $100 million for their space security and readiness technology which allows customers keep an eye on threats to assets they have in space such as satellites. The company’s Jackal autonomous vehicles, which can detect objects in space, are expected to launch aboard SpaceX ‘s Transporter-10 mission next year.

Essential AI, a San Francisco based venture aim to increase productivity by automating time-consuming workflows. They have raised $56 million in funding led by March Capital alongside Advanced Micro Devices, Google and NVIDIA.


All venture data secured from PitchBook


That's all for this week, and this year! Thank you too all 500 of my subscribers. Next year looks set to be a great year for AI, particularly in emerging spaces like AI and MLSec, risk and safety.


Happy holidays and have a great new year!

Mat Holliday

要查看或添加评论,请登录

社区洞察

其他会员也浏览了