Safe, Secure, and Trustworthy Artificial Intelligence

Patent

by Dennis Crouch

The White House this week issued a new executive order focusing on a variety of aspects of regulating artificial intelligence, some of which focuses on IP issues.  The executive order lays out eight guiding principles for manages risks while allowing growth and benefits:

  1. Ensuring AI is safe, secure and trustworthy, including through developing guidelines, standards and best practices, verifying reliability, and managing risks related to national security, critical infrastructure and cybersecurity.
  2. Promoting innovation and competition in AI, such as through public-private partnerships, addressing intellectual property issues in ways that “Protect inventors and creators”, and ensuring market competition and opportunities for small businesses.
  3. Supporting workers affected by AI adoption, including through training, principles for workplace deployment, and analyzing labor market impacts.
  4. Advancing equity and civil rights when using AI in criminal justice, government benefits, hiring, and other areas.
  5. Protecting consumers, patients, passengers and students from risks of AI systems.  “[C]onsumer protections are more important than ever in moments of technological change.”
  6. Safeguarding privacy including through evaluating commercial data use and advancing privacy-enhancing technologies. “[T]he Federal Government will ensure that the collection, use, and retention of data is lawful, is secure, and mitigates privacy and confidentiality risks.”
  7. Improving government use of AI.
  8. Strengthening American leadership abroad to advance international cooperation on AI.

Beyond the stated goals, the order has a number of requirements — most of them directed to the various Federal executive agencies.  The tightest new regulations appear intended to focus on future AI models that are a few times larger than what OpenAI and other are currently deploying as well as “dual-use” models that could have potential national security impact.

  • It requires companies developing or intending to develop “potential dual-use foundation models” to provide information to the government about model training, ownership of model weights, results of AI “red team” testing, and measures taken to meet safety objectives.
  • It authorizes the Secretary of Commerce to define the technical conditions that would trigger these reporting requirements. Until defined, reporting is required for models trained on over 10^26 operations or primarily on biological sequence data over 10^23 operations.
  • NIST is charged with developing standards and tests to ensure that AI systems are safe, secure, and trustworthy and the Department of Commerce to develop guidance for content authentication and watermarking of AI-generated content.

The intellectual property-related aspects of the executive order ask the appropriate agencies to work on the problem.

  • It directs the USPTO Director to issue guidance to patent examiners and applicants on AI and inventorship, including issues related to using generative AI in the inventive process. It calls for the USPTO Director to issue updated guidance on patent eligibility for AI and emerging technologies.
  • It instructs the USPTO Director to consult with the Copyright Office and make recommendations on potential executive actions related to copyright and AI, including the scope of protection for AI-generated works and the use of copyrighted works for AI training.
  • It directs Homeland Security to develop a program to address AI-related intellectual property theft, including investigating incidents and enforcing actions. It also calls for updating the IP enforcement strategic plan to address AI.
  • It encourages the FTC to use its authorities to promote competition in the AI marketplace and protect consumers and workers from related harms.
  • It promotes public-private partnerships on advancing innovation, commercialization and risk-mitigation methods for AI. This includes addressing novel IP questions.

None of these requirements have immediate effect, but indicate that there will be further action over the next few months.

Hide comments

50 thoughts on “Safe, Secure, and Trustworthy Artificial Intelligence”

  1. 9

    Go up 40,000 feet. Computers continue to provide exponentially more processing power. And our understanding of intelligence continues to expand exponentially. The algorithms used now make the ones used in 2005 look ridiculous.

    AI hasn’t quite arrived but with the latest batch of successes we can see that in ten years it will have. All the problems you speak of will be solved.

    The greatest neural scientists in the world thinks that human intelligence is merely the manipulation of latent variables.

    For 101 purposes, there is massive structure in modern AI programs that novel and non-obvious. Much of what is being done now is starting to seem almost magical. And we are only at the start of the great AI revolution.

    Remember that AI is a machine that requires structure to perform functions, space to perform those functions, time to perform those functions, and energy to perform those functions. The conservation of information is one of the most important laws in physics. Computers are transforming information, which requires space, time, and energy.

    Reality. Live in it. It is easier.

    2. 9.2

      “AI is a machine that requires structure to perform functions”

      Describe the structure of a new AI machine in structural terms that distinguish it from the “less intelligent” AI machine that existed one month ago. Note: the new structure needs to be the structure responsible for the improved intelligence, not “structure” that corresponds to mere data content or the meaning of data.

      Good luck! You’ll need a lot of luck because you are one of the least intelligent people commenting here, gramps. History passed you by a long time ago, even before you were wearing those improved Depends undergarments and ranting like a klansman about black presidents.

      1. 9.2.1

        “Describe the structure of a new AI machine in structural terms that distinguish it from“…

        LOL – you do know (as you have been told a million times) that this is not the legal requirement, right?

        Right?

        Right?

        Bueller?

      2. 9.2.2

        “You’ll need a lot of luck because you are …”

        And there he geos again, with his signature
        A
        O
        O
        T
        W
        M
        I (s)

      3. 9.2.3

        >>ranting like a klansman about black presidents.

        Dennis: this is why you need to ban “The Prophet”. He just makes up these defamatory statements with no basis in fact. And they have nothing to do with patents.

        1. 9.2.3.1

          .. or, he can leave them up as clear markers of Malcolm’s instability and people can simply judge Malcolm by his apoplectic rants.

          I am amazed at how little Malcolm can grasp that his choice of rants denigrates himself while he (apparently) thinks that his “mocking” hurts others.

          One would think that those with whom he might share his ideology would chide him, as he makes the Sprint Left crowd look asinine.

  2. 8

    “Half of Cruise’s 400 cars were in San Francisco when the driverless operations were stopped. Those vehicles were supported by a vast operations staff, with 1.5 workers per vehicle. ***The workers intervened to assist the company’s vehicles every 2.5 to five miles***, according to two people familiar with its operations. In other words, they frequently had to do something to remotely control a car after receiving a cellular signal that it was having problems.”

    Just a regular reminder that some commenters here have pointed to SF as an example of how great the “self driving” (LOL) car experiment was going. The imminent life-improving glories of so-called “artificial intelligence” are being peddled by the same con artists (and rube mouthpieces). Guess how it’s going to play out?

    1. 8.1

      Keep on showing your 1gn0rance of how innovation often in its initial stages is actually worse than the status quo.

      You really need to get a c1ue.

      1. 8.1.1

        “innovation often in its initial stages is actually worse than the status quo”

        That would explain the low prices. LOL

        In any event, tell it to the tech peddlers who made so many ridiculous (and later proven FALSE) claims about “self-driving” vehicles. Those claims were, in turn, either swallowed whole by many of the self-identifying “tech savvy” commenters here or allowed by other folks to pass without noting their obvious absurdity.

        1. 8.1.1.1

          And yet again, always feel free to abstain from the computing innovations that you would deign not to have innovation protection for.

          (Like that would ever happen… )

    2. 8.2

      I think I remember reading somewhere that the first motorcars were required somewhere to be proceeded by a man walking with a red flag to avoid terrorizing horses?
      More seriously, the need for so much human supervision of driverless cars these days illustrates the huge difference between using AI on physical objects moving in the real world, versus merely analyzing and manipulating digital data. The former has to deal with various physical sensors and their limitations and reliabilities, plus deal with many unusual situations not ever yet even programmed into the AI control software. [Fire hoses or car reck parts in the road, etc., etc., etc.] Robots can also have such issues. There is a good reason why many big factory robots are in locked steel mesh cages.
      Also, a valid safety comparison has to be made vis a vis the Average human car driver, and that safety record is statistically worse than that of even front line soldiers in many wars.

      1. 8.2.1

        Making a ride “safer” for the occupant of the computerized car is actually the EASIER part of computerized driving. Start by piloting the vehicle as slow as legally permitted, be extra, extra cautious and very liberal with braking, and only use the routes with minimal traffic and turns. The problem with that is (1) hardly anyone wants to buy a vehicle like that and (2) hardly anyone wants to share the road with a vehicle like that.

        The real dream of the self-driving car peddlers, of course, is to force everyone into their self-driving cars by banning the “dangerous” and “obsolete” human-operated cars. That’s how the Silly Con Valley bros operate and it’s a stunt they will never tire of (pun intended).

        By the way, if actual automobile-related safety improvements were important to anyone in power, we’d see a number of incredibly obvious changes to dashboards and vehicle design that have nothing to do with computers or “innovation” in any field.

      2. 8.2.2

        Paul, go up 40,000 feet. Computers continue to provide exponentially more processing power. And our understanding of intelligence continues to expand exponentially. The algorithms used now make the ones used in 2005 look ridiculous.

        AI hasn’t quite arrived but with the latest batch of successes we can see that in ten years it will have. All the problems you speak of will be solved.

        The greatest neural scientists in the world thinks that human intelligence is merely the manipulation of latent variables.

        For 101 purposes, there is massive structure in modern AI programs that novel and non-obvious. Much of what is being done now is starting to seem almost magical. And we are only at the start of the great AI revolution.

        Remember that AI is a machine that requires structure to perform functions, space to perform those functions, time to perform those functions, and energy to perform those functions. The conservation of information is one of the most important laws in physics. Computers are transforming information, which requires space, time, and energy.

        Reality. Live in it. It is easier.

        1. 8.2.2.1

          Re: in ten years “All the problems you speak of will be solved.”
          Like the software hard-crashes on my new MS 11 PC that I still have occasionally even though MS has had more than 30 years to work on it and I am a careful user?
          Even though that kind of software does not even require all the physical sensors and myriad different physical world encounters that I was talking about?

  4. 6

    The current fervor over AI reminds me much of the nanotechnology boom almost 20 years ago — Lots of huge predictions about how its going to change everything. While I don’t doubt nanotechnology is still with us, particularly in the continued miniaturization of electronics, I don’t believe it is the game-changer that it was hyped to be, and is much less talked about today. The tone, tenor, and depth of must AI treatment in mass media strikes me in much the same way now as that did then.

    1. 6.1

      The “AI” fervor is worse than that because at least with “nanotechnology” there was some shared understanding of what exactly was being peddled. Also, because that tech was typically grounded in the (small) physical world, people could evaluate the claims being made without getting caught in a vapor pit. Hence the fall of the con artist and disgusting fraud Elizabeth Holmes.

      The folks pushing “AI” are going to do a lot more damage before the curtain is pulled back. Some of the damage will be intentional but a lot of it will flow from the fact that the AI peddlers are just incredibly greedy, fantastically unintelligent narcissist parasites.

      1. 4.1.1

        Nope. The big corporations want it to cost you at least $1 million dollars to comply with the federal regulations so that only big corporations can use AI.

        Just f’ing unbelievable.

        1. 4.1.1.1

          Setting up another Kings game, where only the large companies can win under the regulatory scheme, and of course sandbag everyone else especially those rascal start-ups/potential disruptors. Same as it ever was. So therein emerged a new technology, and it took what 20 months ? before DC stuffed that puppy in a bag and gave big tech bats to swing.

          Just wait for the regulations. LOL.

  7. 3

    Supreme Court observers seem to think, from the oral arguments Wednesday, that most justices were skeptical of the right to trademark the phrase “Trump Too Small” for use on T-shirts. Chief Justice Roberts apparently even wondered if it could make it harder for others to make their own T-shirt takes on Trump?
    Whisky bottle dog chews, and now this? Why is it that so many of the so few IPL cases that make it to the Supremes seem to have such IPL-profession unflattering facts? At least they apparently got it straight that this is not any free speech restriction.

    1. 3.1

      “ At least they apparently got it straight that this is not any free speech restriction.”

      Baby steps.

      1. 3.1.1

        ?

        Your prior position and this post is beyond murky (not a surprise, given how badly you understand 1A in view of patents).

        1. 3.1.1.1

          My position is that denying trademarks is not a free speech restriction. Also you’re a barely conscious derpwad (TM). I’ve held these positions consistently.

          1. 3.1.1.1.1

            The response after yours contains the case that makes your stated positions to be inconsistent: Tam.

            Of course, your tendencies towards One-Bucketing and Ends Justify the Means means that you are self-deluding when it comes to your hypocrisies.

            But you be you.

            1. 3.1.1.1.1.1

              My consistent position on Tam is that Tam was decided incorrectly. It made no sense and if the Supremes effectively nullify the Tam decision with a decision that actually makes sense (as they did with their similarly senseless decision in Diehr) then that would be an excellent result.

              I look forward to the same thing happening with the Court’s recent appalling jurisprudence relating to women’s rights, although I would prefer for the misogynist dominionist lizards on the Court to perish in a fiery wreck today, if possible.

              1. 3.1.1.1.1.1.1

                You were off then. And that does not excuse your inconsistency with the present case.

                As for “women’s right’s,” is that only women fetuses, or does that include male fetuses?

                And — by the by – YOUR “Ends Justify the Means” views in patent law were exactly the driver of what you get outside of patent law — expressly predicted by me (along with your reaction).

                Yet again, Malcolm – I have the far better positions.

                1. …

                  “women fetuses”

                  Very serious stuff here from one of this blog’s most eligible bachelors. LOL

      1. 3.2.1

        Yes, the Supremes will be dealing with their 8-0 Jun 19, 2017 decision in Mattel v. Tam re his band, “The Slants,” holding that the first amendment overrides a TM statute prohibiting defamatory-mark registrations.

      2. 3.2.2

        I am wondering how such a hypothetical position could possibly be squared with Tam.

        Ditto. I was surprised to read how much of the argument focused on whether it even is a free-speech violation to deny registration. I was never convinced that denying TM registration should be regarded as a free-speech violation, but Tam said that it is.

        I could understand how one might draw a distinction between this case and Tam, in that the party “injured” by the registration of “SLANTS” is an amorphous collective, while the party “injured” by “TRUMP TOO SMALL” is a specific individual. Maybe that makes the strict-scrutiny balance work out differently.

        I am baffled, however, as to how one can conclude that it is a burden on speech to deny registration to “SLANTS,” but not any sort of burden at all to deny registration to “TRUMP TOO SMALL.” If it is a burden for one, then surely it is a burden for the other.

  8. 2

    Everyone who believes our could-use-some-AI-himself president actually understands even half of what his advisors spoon-fed and wrote for him, please raise your hand . . .

    At a time when our country — and indeed the world — could use a John Kennedy or a Ronald Reagan, are Biden and Trump really the best we can do?

    Really?

    1. 2.1

      “ At a time when our country — and indeed the world — could use a John Kennedy or a Ronald Reagan”

      LOLOLOLOLOLOL

      Parents, please don’t let your nine year olds play on the Internet.

    2. 2.2

      Elizabeth Warren would make a fine POTUS with Adam Schiff as VP.

      Trump will choose Mike Johnson as VP for the same reason Bush Sr chose Quayle as VP. It will make him impeachment-proof.

Comments are closed.