Welcome to the Future of Life Institute newsletter. Every month, we bring 29,000+ subscribers the latest news on how emerging technologies are transforming our world - for better and worse.

If you've found this newsletter helpful, why not tell your friends, family, and colleagues to subscribe here?

Today's newsletter is a 10-minute read. We cover:

• Our new short film, “Artificial Escalation”

• Actors and writers pushing back against job automation

• FLI cause areas in recent media

• The UN Security Council and the US Senate discuss AI

Our new short film, “Artificial Escalation”, explores this very real possibility, demonstrating a global catastrophe driven by the following factors:

• Accidental conflict escalation at machine speeds;

• Al integrated too deeply into high-stakes functions;

• Humans giving away too much control to Al;

• Humans unable to tell what is real and what is fake, and;

• An arms race that ultimately has only losers.

But this doesn’t have to be our fate. If you haven’t seen it already, be sure to watch below and share. You can find more information on our work to keep AI out of NC3 systems at the link here.

Nuclear weapons, artificial intelligence, and autonomous weapons seem to be getting more screen time than ever before.

• Most notably, Oppenheimer premiered in cinemas, recounting the life of the so-called “father of the atomic bomb”. It couldn’t be more relevant than it is now. The Bulletin of the Atomic Scientists warns of the great potential for imminent global human-made catastrophe, such as a nuclear exchange. And as director Christopher Nolan pointed out on a recent panel, we’re in an “Oppenheimer moment” for AI, in the risks it presents on its own, and the danger from its potential integration into NC3.

• The newest Mission: Impossible film focuses on the existential threat presented by a mysterious, artificially-intelligent cyber villain; its “most topical villain yet” according to one columnist. Although fictional and not representative of current AI systems, it’s a relevant cautionary tale for our time.

• On Netflix, the Unknown: Killer Robots documentary (featuring FLI’s Dr. Emilia Javorsky) presented a terrifying reality with autonomous weapons. One reviewer wrote after seeing it, “the future of AI will fill you with unholy terror”.

AI policy:

• The U.S. Senate held a hearing this week on “Oversight of AI: Principles for Regulation”, with testimony from AI experts Stuart Russell, Yoshua Bengio, and Anthropic CEO Dario Amodei. The three outlined concerns about AI risks and our ability to mitigate them, as companies continue to develop AI tech at an unforeseen, "worrisome" speed. Read our statement on it here.

• The United Nations Security Council held its first meeting on AI on 18 July, with Secretary General António Guterres urging for “transparency, accountability, and oversight”.

Autonomous weapons:

• In the same UN Security Council meeting, Sec. Gen. Guterres called for a legally binding instrument on autonomous weapons to be negotiated by 2026. He also spoke of the need to maintain human agency and control over nuclear weapons.

Climate change:

• The G20 environment meeting in Chennai, India ended on Friday with no agreement on climate policy - a disappointing result following a month of intense, record-breaking heat.

• AI pioneers Stuart Russell and Geoffrey Hinton, with FLI’s EU Research Lead Risto Uuk, met with European Commission EVP Margrethe Vestager to discuss AI safety.

• Earlier this month, we announced our first nuclear war research grants. This grant round supports 10 projects examining the humanitarian impacts of nuclear war. More information on our 2023 grants can be found here.

• On the FLI podcast, guest host Nathan Labenz spoke to Skype and FLI co-founder Jaan Tallinn about his views on AI development.

• Also on the FLI podcast, host Gus Docker spoke to writer and Roots of Progress founder Jason Crawford about Jason’s philosophy of progress, looking back into history and projecting into the future.

Automated jailbreaking: New research from Carnegie Mellon University and the Center for AI Safety exposes the vulnerability of large language models (LLMs) to adversarial attacks that can bypass their safety measures, with no known way to prevent such attacks.

Why this matters: The ability to avoid safety guardrails in widely-available LLMs like ChatGPT leaves the door open for anyone to misuse them, for example by having an LLM generate the harmful content it’s supposed to be aligned against distributing. It also ties in with the ongoing open source vs. closed-source debate in the AI community, as the attacks studied by the researchers seem to impact both kinds of LLM.

• “AI Bioconvergence”: Helena recently released their report on AI-enabled biology, examining the intersection of AI and biotechnology, and its potential impacts on biosecurity.

• A new Turing test: In the MIT Technology Review, Mustafa Suleyman proposes a new way to measure the “intelligence” of AI - can it make $1 million?

• Before (or after) you see Oppenheimer…: We recommend you read through, and share, this supplementary fact sheet produced by the International Campaign to Abolish Nuclear Weapons.

• What we’re watching: Kurzgesagt’s short animation on bio-risk from earlier this month provides a great introduction to the topic.

Shortly before the release of Oppenheimer, 16 July marked the 78th anniversary of the Trinity test - the first detonation of a nuclear bomb.

The Trinity test effectively began the nuclear age, making tangible the unbelievable power to destroy that nuclear weapons introduced to humanity.

The effects of the blast, which sent a radioactive mushroom cloud high into the atmosphere, weren’t fully known until years after. A recent report suggests that its radioactive fallout made it to 46 U.S. states, along with Mexico and Canada. The consequences of this were felt most in the New Mexican communities closest to the test site; infant mortality rose 56% in New Mexico in the three months following the test, and so-called “downwinders” suffered from elevated rates of cancer for decades to come.

Trinity was only the first test of hundreds to come, and the first nuclear bomb of the estimated 13,000 that exist in the world now - many of which are far more powerful than Trinity. There’s no question that nuclear weapons are devastating, and a nuclear exchange today would be inconceivably catastrophic for a majority of the world. However, we don’t have to accept these risks.

Read more on our website here about our work in this space, and the actions necessary to safeguard our future.