Will he, or won’t he?

That’s the big question right now when it comes down to whether Governor Newsom will sign the California Senate Bill 1047 entitled the “Safe and Secure Innovation for Frontier Artificial Intelligence Models Act”.

SB 1047 is variously known as the AI Safety Act, the AI Existential Risk Act, and goes by other informal designations. The upshot is that this key piece of AI-focused legislation has garnered outsized attention throughout the nation because of its sweeping attempt to encompass legally binding AI governance constraints and would be the first of its kind if it indeed goes into law.

A lot is at stake.

Right now, the pins and needles circumstance is that the controversial bill sits on Governor Newsom’s desk and awaits his action, which must be taken by September 30, 2024. The clock is ticking. All eyes are on him at this juncture. He can either approve SB 1047 by signing the legislation into enactment, he can veto it or some say kill it, or he can do nothing which by default will cause it to be enacted via a provision known as a pocket signature.

Per ongoing coverage in my Forbes.com column about the latest in AI including ethical and legal ramifications (see the link here), this discussion will explain what SB 1047 is about and provide insights regarding the tradeoffs associated with approving versus vetoing the bill. I will endeavor to be succinct. Please keep in mind that there is a myriad of complexities involved, and I wholeheartedly encourage you to look further into the matter if interested.

What Is Being Said About SB 1047

There are seemingly as many powerful forces lined up against SB 1047 as there are in favor of the bill. AI luminaires of all kinds have made their opinions known, quite fervently. You can readily find this or that AI pioneer proclaiming that it is the best thing since sliced bread and yet in the same breath find other equally notable AI forerunners that loudly decry the legislation as untenable. The heated debate keeps going back and forth can be confounding as to who is right and who is wrong.

I’ve tried to boil down some key position points into a set of eight especially vital considerations by each respective side of the contentious matter.

Here are some of the key position points in favor of SB 1047:

  1. Seeks to prevent catastrophic harm from AI.
  2. Legally puts both AI makers and AI deployers on the hook for their AI.
  3. Stipulates that a semblance of reasonable care is required about AI.
  4. Provisions for emergency shutdown or kill switch of AI.
  5. Serves as a role model approach for other states and the federal government.
  6. Time is of the essence and must be enacted now before AI gets further out of hand.
  7. The societal benefits far outweigh the societal costs.
  8. If not enacted there will be doom and gloom.

Here are some of the position points opposed to SB 1047:

  1. Unreasonably broad and egregious government overreach on AI.
  2. Will inhibit innovation and curb AI advancement.
  3. Overburdens with excessive compliance and bureaucracy.
  4. Impractical at the get-go and a misguided way to proceed with legislating AI.
  5. Will spur unfounded and draining legal wars against AI firms and AI deployers.
  6. Premature and jumps the gun on what’s needed about AI.
  7. The societal costs far outweigh the societal benefits.
  8. If enacted there will be doom and gloom.

You might keenly observe that both sides tend to assert that there will be doom and gloom if the opposing side prevails. Those who favor enactment argue that humankind will be at the mercy of AI if the bill isn’t put into law. Those who oppose enactment argue that the AI industry in the US will stop innovating and the country will be at a substantial disadvantage to countries that are eagerly proceeding ahead on AI advancements.

Head spinning, for sure.

AI Bills Aplenty In California

At a major tech conference last week, during an annual event known as Dreamforce put on by Salesforce, Governor Newsom signed some pending AI bills while on stage, doing so on September 17, 2024. There was SB 926 which makes it now a crime in California to blackmail using AI-generated nude images, SB 981 which makes social media provide for reporting of AI-produced deepfake nudes, and SB 942 which requires generative AI systems to disclose AI-generated content. There are around thirty other AI-related bills also pending his decision to sign or not.

Regarding the status of SB 1047, Governor Newsom said this while onstage: “There’s one bill that is sort of outsized in terms of public discourse and consciousness; it’s this SB 1047.” Continuing, he stated “What are the demonstrable risks in AI and what are the hypothetical risks? I can’t solve for everything. What can we solve for? And so that’s the approach we’re taking across the spectrum on this.”

All in all, the news media noted widely that those remarks seemed to suggest that SB 1047 might not get approved and that this was a venue and moment to signal what might be coming. Perhaps an early indicator to get those in favor of the bill to realize that things might not go in their direction. It could be likened to forewarning before an earthquake hit, a California kind of signaling.

Why SB 1047 Might Not Make It

Let’s read the tea leaves as to why a veto might indeed be in the offing.

  • AI luminaries are at loggerheads about SB 1047.

Since well-known AI experts are lined up on both sides of SB 1047, there is no viable means to declare that AI insiders are entirely on one side or the other. This means that it is a toss-up when it comes to asserting that AI specialists either favor or oppose SB 1047 per se. The matter is ostensibly at a tie concerning expert authoritative testimonials.

  • Major AI tech firms oppose SB 1047.

A bevy of major AI tech firms oppose SB 1047. Those firms are highly influential and can be instrumental in future stakes and aspirations regarding the impacts of whether the bill gets signed or is vetoed. Enough said.

  • Nancy Pelosi opposes SB 1047.

When the California legislature passed SB 1047, Nancy Pelosi came out in opposition of the bill as noted in a written statement entitled “Pelosi Statement in Opposition to California Senate Bill 1047” that said this: “The view of many of us in Congress is that SB 1047 is well-intentioned but ill-informed.” And further noted, “While we want California to lead in AI in a way that protects consumers, data, intellectual property and more, SB 1047 is more harmful than helpful in that pursuit.” (Source: Official Press Release, Congresswomen Nancy Pelosi, California’s 11th District, August 16, 2024). Indubitably, this is a rather prominent opposition that undoubtedly weighs into the calculus of signing versus vetoing SB 1047.

  • Lots of other AI bills in California got signed and still more might be signed.

In the minds of those in the know, SB 1047 is the elephant in the room, the 600-pound gorilla, the whale of an AI bill. That being said, the public at large likely is unaware of the outsized nature of the legislation. As noted, various AI bills have already been signed in California, and many more are waiting to be signed and will most likely many will get signed. The crux is that there is plenty of opportunity to emphasize that all manner of AI laws were enacted this year. Touting this could aid in dealing with any potential blowback about SB 1047 not making the cut.

  • The California legislature can presumably hold its head high on AI legislation.

Suppose SB 1047 gets vetoed. The California legislature can nonetheless hold its head high and say that it did what was needed to push forward on this kind of AI legislation. The buck stopped elsewhere.

  • Easy blame game on the feds.

One issue that has been heavily debated is whether Congress ought to devise and enact this kind of AI legislation, thus arising at the federal level and essentially providing a role model for the states and avoiding a possible messy state-by-state approach. Congressional bills have been proposed but didn’t make it out the gate. You can bet that none will pass until at least after the upcoming November elections. Realizing that September 30, 2024, is going to arrive before those elections, there is nearly zero chance of a fed AI bill magically getting passed. In short, it is possible to readily blame the feds for not taking action and thus veto the California SB 1047 with the pronouncement that this is something Congress ought to take on.

  • Kicking the can down the road seems an option

First, note that the other AI legislation that has passed in California tends to emphasize the here and now and speaks to everyday qualms about AI. SB 1047 is more so about AI as an existential risk. It is principally future-based rather than purely immediate-based. Sure, some say we need to take action right away, but others assert that this would be premature and stifle AI innovation. If SB 1047 gets vetoed at this time, this doesn’t somehow stop the legislative pursuit in general. Once the results of the November elections are known, it might be quite straightforward to devise another similar bill and move it along again to the governor’s desk. Right time and place, as it were.

  • Governor Newsom has made signals.

As a reminder, the remarks noted above that Governor Newsom made at the recent tech conference would seem to be the proverbial canary in the cage. You could suggest the comments were intended to soften the blow of a potential veto. Start getting everyone ready for the weighty matter. This is likely more astute than seemingly out-of-the-blue summarily vetoing SB 1047 and leaving those who staunchly favor it in an utter state of shock. Maybe additional signals might occur this week ahead. Keep your antennas up.

What Will Happen Next

You can bandy around the status of SB 1047 until the cows come home. Well, actually, that’s going to be no later than September 30, 2024. Set your calendars accordingly.

It would seem unlikely that a pocket signature is going to occur. That would be if Governor Newsom takes no overt action, and the bill automatically is enacted. This wouldn’t avoid an uproar by those opposed to the bill. He would still get pinned, regardless of default approval or actively signing it.

The rock and hard place is that pretty much the likely choice is either to sign SB 1047 or veto SB 1047. Life is tough sometimes.

If SB 1047 is going to be signed, presumably doing so at any time between now and September 30 is about the same. To make a bigger splash, perhaps Monday, September 30 would be good, assuming that there aren’t some other newsy activities expected on that day. For a veto, if the idea is to keep this low-key, a late afternoon action in California on Friday, September 27, 2024, would be handy, occurring after the news cycle on the East Coast has subsided for the day and going into a weekend.

Right now, we can only just wait and see.

Will he, or won’t he?

Time will soon tell.

Share.
Exit mobile version