Paul Nakasone Joins OpenAI’s Board of Directors
Former NSA Director Paul Nakasone has joined the board of OpenAI.
Comments
Winter • June 24, 2024 8:27 AM
I don’t want to sound sarcastic, but I do not see how having a former NSA director as a board member will increase the trust of OpenAI’s (international) clients and users in their own security.
vaadu • June 24, 2024 9:58 AM
Edward Snowden says OpenAI just performed a “calculated betrayal of the rights of every person on earth”
“they’ve gone full mask off: do not ever trust openai or its products.”
Logan • June 24, 2024 10:14 AM
Do you think someone who worked for the NSA, a job in which you hold some of the most confidential and important information, should be working with OpenAI, whose sole purpose is to feed information into an AI to make it as smart as possible?
Who? • June 24, 2024 10:39 AM
Now we know what goal OpenAI has. Just in case worldcoin was not clear enough, I guess.
yet another bruce • June 24, 2024 12:04 PM
Makes sense to me. OpenAI and the NSA are two organizations founded on the insight that “Attention is All You Need”
Clive Robinson • June 24, 2024 12:05 PM
The opener to the article claims,
“… as a tough regulatory environment pushes tech companies to board members with military expertise.”
Really?
But look a little further down and you find,
“Until January, OpenAI had a ban on the use of its products for “military and warfare.” The company says the prohibition was removed to allow for military uses that align with its values, “
Now that is an area that has money sloshing around by the barrel.
Which is why towards the bottom you find,
“Nakasone brings deep Washington experience to the board, as the company tries to build a more sophisticated government relations strategy and push the message to policymakers that U.S. AI companies are a bulwark against China.”
The US-v-China is a political boon dongle used to throw money around and dig up the same nonsense that drove the “Una-American activities” and “Reds under the bed” fear mongering for profit and political gain.
‘https://en.m.wikipedia.org/wiki/House_Un-American_Activities_Committee
The current version is called by some “The China Scare” and is a continuation of retorhic that apparently never goes away but resurfaces in waves since the mid 1800’s about the nations around the South China Seas and across into India and as far as the “Stans”.
It is often seen as the basis of George Orwell’s book “1984” use of a distant enemy called “Eastasia” and “little yellow men” political propaganda.
Which is what this article comes across as.
But what of A.I.? Well quite a few realise that the Orwellian surveillance state is not possible with just human resources. Places like East Germany via their history show us that means other than surveillance have to be used, as do a number of south American and other “Police States”. Many of which have failed in one way or another.
Some now think rightly or wrongly that the main use of “future AI” will be as a method of “Universal Surveillance” and the creation of “Enemies of the State” some of which we already see with scandals about “RoboDebt” and what went on in the UK “Post Office”.
My view is that the current LLM and ML, AI systems are already being seen as a way to increase surveillance and control on ordinary people. I seriously expect that current AI systems will get used in the US as a form of “Voter Deselection” system the intent being much like that we see every voter season with the likes of Gerrymandering and “similar name exclusion”.
It will be interesting to see what happens later this year and in a couple of years time, but I suspect voter deselection and similar will be found to be in use.
Boardmember • June 24, 2024 7:17 PM
What exactly is the purpose of a board? Is it to be a check on the executives of the organization or is it to be enablers of those executives?
One could argue that people like Nakasone are useful to have on hand for when you want to go after government contracts because of the experience they bring (leaving aside the argument of the government revolving door). That, however, does not mean that they should sit on your board, that means they should be advisors or consultants.
Putting these people on your board makes a mockery of what a board should occupy itself with. While this is no surprise to anyone (here or elsewhere): OpenAI is just another digital pillager like most of Silicon Valley, looking out for no-one but itself.
Clive Robinson • June 24, 2024 10:42 PM
Hot off the press…
Re : US v. China political boon dongle.
“WASHINGTON, June 24 (Reuters) – The Biden administration is investigating China Mobile, China Telecom and China Unicom over concerns the firms could exploit access to American data through their U.S. cloud and internet businesses by providing it to Beijing, three sources familiar with the matter said.
Authorities at the Commerce Department are running the investigation, which has not been previously reported. They have subpoenaed the state-backed companies and have completed “risk-based analyses” of China Mobile (0941.HK), opens new tab and China Telecom (0728.HK), opens new tab, but are not as advanced in their probe of China Unicom (0762.HK), opens new tab, the people said, declining to be named because the probe is not public.”
This is further evidence that the US Adminstration has placed the cross-hairs on anything “Chinese” in technology to destabilise world communications and the economics that it supports.
It’s the next steps in the next stage of pushing not just China but all other Nations to use US tech and suffer what that entails in the way of surveillance.
It’s the next generation legacy of a failed policy that goes back to the 1970’s and ended up with the World not using “US-Telco Tech” and GSM becoming the predominant mobile communications technology world wide.
What is not being talked about very much at the moment is that the next generation of telecommunications to replace the back-haul infrastructure is going to be based on AI technology. That is all the things you hear about such as “Signalling System Seven”(SS7) / “Common Channel Signaling System 7″(CCSS7),
‘https://www.techtarget.com/searchnetworking/definition/Signaling-System-7
That are responsible for much of the signal routing, connection, monitoring and management are going to get “blessed by AI” if certain US Corps have their way (Google in particular are “quietly” behind this initiative as there is considerable US Gov military style spending behind it, of the sort that as we know has previously upset Google Employees which is why they try to make it look like humanitarian technology ‘https://www.extremetech.com/defense/google-will-give-the-military-ai-assistance-when-disaster-strikes ).
The first public indication of what was being started as I’ve indicated before was back at the UN ITU Conference in Doha back in 2014. It had become clear that certain “power blocks” were deeply unhappy about US control of not just the Internet but much of all communications infrastructure that were forced through “Choke Points” under the countries that formed the Extended Five-Eyes. Since then the likes of Google, Amazon, SpaceX and similar have been pushing their way into the international data communications back bone at over a hundred nations boarders.
Part of this that is not widely known has been that they all have been “contracted” in various ways by the United States Department of Defense. Most obviously is “Starshield”, which is rather more than just a military version of Starlink and is designed for “US government use” to not just control other nations communications but carry out even more than communications Surveillance.
As was noted a couple of years back by Elon Musk, SpaceX can put up –the upto 35 thousand– Starlink satellites way faster and for way less cost than Putin can shoot them down with current ASAT technology.
The use of AI to control other nations communications is very high on the US Mil/SigInt etc agencies list of desirables, even more so than certain types of “drone tech” as it gives deniable soft power projection rather than hard that the likes of drones and sixth gen mil aircraft unloading munitions and at a fraction of the cost.
Clive Robinson • June 24, 2024 11:05 PM
@ Bruce, ALL,
Re : UN veiled threat at AI etc Corps
Citing “climate change” which AI is a major contributor to more so than the Social Media networks it’s now being put behind
United Nations Secretary-General Antonio Guterres demanded on Monday that major consumer technology firms not just take responsibility for the significant harm they are doing but also,
“Acknowledge the damage your products are inflicting on people and communities.”
And,
“Some stakeholders carry an outsized responsibility,” he added. “For them, I have a clear message: We demand action.”
Clive Robinson • June 25, 2024 5:59 AM
My view is that the current LLM and ML, AI systems are already being seen as a way to increase surveillance and control on ordinary people. I seriously expect that current AI systems will get used in the US as a form of “Voter Deselection” system the intent being much like that we see every voter season with the likes of Gerrymandering and “similar name exclusion”.
It will be interesting to see what happens later this year and in a couple of years time, but I suspect voter deselection and similar will be found to be in use.
JonKnowsNothing • June 25, 2024 11:19 AM
@Boardmember , All
re: What exactly is the purpose of a board
(USA)
There are a few items on the list of what board members do, but it depends on the size of the company. In this case this is a new, very large, expected to be wildly profitable company with many links already to Venture Capital and The Big Dogs of computing.
The primary purpose of a board member in this case is to
- Maximize Shareholder Profits and Returns
Companies do not need to be listed on the stock exchanges for this purpose, as the shares are privately held but profits will be distributed according to internal methods (which vary).
Members like Nakasone are compensated for their time and meetings, for which they can get sizeable payments but also with a huge pile of stock or stock options. Theoretically this allocation keeps the company’s interests forefront because the board members have skin in the game. This rarely is the real reason but looks good on paper.
The day to day running of a company is the responsibility of the executives of the company; the board has little to do with that.
The board is intended to have periodic reviews of major corporate changes and given the size of their stock holdings can have impact on major alterations. In a proxy fight stock holdings determine the winner(s).
- The last time the OpenAI board differed with the executives; SAltman was canned and a few days later was returned with full honors and benefits. All the opposing board members were terminated.
When people who held major positions in business or government “retire”, they get on The Board Member Gravy Train. They are not going to get a job at Amz hauling 40lbs of cat litter 10hrs a day. They get on multiple boards of a nice group of companies.
Free travel, nice destinations, good food, money to boot. It’s a good gig.
- You are not invited
Paul Lock • June 25, 2024 12:30 PM
There is no doubt about the concerns we all have, both about the risks of badly designed technology, and about intentions of badly motivated executives.
Please remember the fastest bulwark against both problems in this instance is the protection of copyright and privacy rights. I hope everyone considers the role they can play in bringing fair and effective legislation into being.
The legal framework that grew to maintain confidence in the economy in the last century (contract law, executive liability, shareholders’ rights, protecting against fraud, transparency, requiring timeliness, libel, and consumer protections), should be quickly used as the template to protect all aspects of democratic and civic life from the negative human behaviors that are common to both worlds.
p.s. Bruce, It pains me greatly to see that even your lovely garden now requires moderation.
Regards,
Paul Lock
GCHRD.org
Clive Robinson • June 25, 2024 12:37 PM
@ALL
The C-38 was based on an additive “coin counting” mechanism originally designed for vending machines (think cigarette / chocolate bar machines).
It was known to have a very wide spread on “Key Strengths” with only about 15% being strong and 20% being weak.
These key strength issues were known to William Friedman and seen as desirable.
Put simply it was known that these cipher machines being used in tactical / frontline roles would fall into enemy hands and be easily duplicated.
So having the “keys” centrally issued ment that “strong” keys would be used, yet an enemy using the machines would without knowing about the key strength issues would use a mixture of strengths.
The thing is that the use of a weak key gives a lot of information that can then be carried forward to attack other messages via “probable text” etc.
This trick was carried forward into all the Crypto AG cipher machines sold around the world.
It was only with the advent of entirely electronic cipher systems with suitable anti-tamper etc that the non-civilian use cipher systems become suitable for tactical use.
jg865kh • June 25, 2024 12:51 PM
I started to write a long-winded post on the parallels between Condoleeza Rices’ appointment to the Dropbox board of directors in 2014 and Paul Nakasone’s appointment to the OpenAI board of directors today but realized it doesn’t really matter all that much.
Then as now, it’s just to get a government friendly member on the board of directors. Then as now there is “public outcry” that will be patronized and then ignored going forward. Then as now it only benefits the company and will provide no benefit to the people who are being monetized.
And then as now there will always be a few of us that will avoid Dropbox and OpenAI and maybe support each other in finding ways to do so
I look forward to the discussions that follow, perhaps on Bruce’s blog, on how to minimize exposure to OpenAI or any AI for that matter, after the outcry subsides.
Wat • June 25, 2024 12:56 PM
Given that the military has been crowing for years about their lack of AI talent and the need for AI-fueled capabilities and how China has AI-guided targeting for military equipment and if we don’t match them then we’re at a military disadvantage etc etc….
It seems pretty clear-cut to me that the purpose of Nakasone joining OpenAI’s board is to convince them to invest more of their resources into studying and developing applications of AI that have dual-use that the military can use (or even single-use for only military use).
Beyond that, I’m not sure just how much Nakasone’s presence would actually help OpenAI get DoD contracts. General Alexander’s ill-fated IronNet security went belly-up due to that former NSA chief’s inability to power through the hurdles posed by DoD acquisition law; would OpenAI face the same issue? Or are the economics here so wildly different that the barriers faced by IronNet not applicable to OpenAI?
Ray Dillinger • June 25, 2024 1:17 PM
I wish I had something insightful to add but it really doesn’t seem like there’s any subtlety to penetrate here. This is pretty clearly exactly what it looks like. OpenAI has been co-opted.
On the one hand, we have ongoing covert influence campaigns from troll farms foreign and domestic. Some are national, from countries or organizations opposed to the USA itself. Others are after money or power and completely don’t give a shit about the USA government except insofar as it’s an obstacle to their interests. Monitoring and stopping those is a legit need.
But the more tightly tied to any political interests the people doing it become, the more likely it is that they represent one of those attacking groups that are after money or power rather than any effort to stop them.
And we have people in this nation specifically talking about their plans to purge all the “disloyal” people from the security and police agencies and make them subordinate to the executive branch under one political party, which guarantees that the takeover of OpenAI will be exactly that kind of attack.
Clive Robinson • June 25, 2024 11:19 PM
@Moderator,
Some of the comments just put up were clearly not for this thread but the thread on the C-38 / M-209.
JonKnowsNothing • June 26, 2024 1:56 AM
@All
re: Influence Peddling
Influence Peddling is a form of corruption however, it has to take place within a very defined set of conditions. Those conditions are determined by jurisdiction and country. What is normal business practice in one place is not the same in another.
Consider: tips (aka gratuity)
- TIP: To Insure Promptness
Depending on who gets the funds it may be normal business exchange or an illegal action
- It is illegal to offer tips to some groups of workers, such as U.S. government workers and more widely police officers, as the tips may be regarded as bribery.
However, in the USA it is customary to give a tip to someone who brings food to your table. In Europe, tips are often not offered as they are included in the Prix Fixe of the meal. Europeans are getting the hang that Americans will drop an extra 20%-50% even when it’s not required. The practice varies by country, region and activity.
So, something as simple as a tip can be both legal and illegal.
In the circumstances of influence peddling in corporations, the legal team is very keen not to cross a legal threshold. They do slip up and we get splash media cases when it happens.
Any influence peddling will be done under careful conditions. No one would believe it doesn’t happen, but in order for the action to become a legal issue it has to meet jurisdictional criteria. This is not an easy threshold to document legally, although we know it when we see it (or hear about it).
Consider: HP v Autonomy
A case that took from 2011 to 2024. It was alleged that serious fraud and account issues existed with the company and HP did not realize the accounting was in error until after the purchase was completed. Billions of USD changed hands. A large number of people were not happy. Nasty legal exchanges and criminal complaints were filed.
In the end, all that bluster didn’t meet the requirements.
- The court heard evidence and arguments over the course of 11 weeks, with the number of charges being reduced from 16 to 15 as one count of securities fraud was dropped. The jury retired for deliberation on 4 June 2024. On 6 June 2024, he was found not guilty of all charges.
It may look like a duck and walk like a duck but it’s not a duck until the court says it’s a duck.
Clive Robinson • June 26, 2024 10:27 AM
@ All,
Re : AI attack and defence.
The “on boarding” of military / government means one thing for sure, there is going to be an active marketplace develop in vulnerabilities and attacks on AI LLM and ML systems.
Yes it was a foregone conclusion much like water drips downwards but so far the consequences have actually been more academic than significant, though that is about to change fairly big style over the next twelve months at most.
LLM’s and ML AI are to be blunt very vulnerable and it’s fundamental to not just their functioning or design but the fundamental way they work.
Which is why the likes of Nicholas Carlini and others are finding so many vulnerabilities they are more than busy.
However we are seeing “NIH and Golden Goose” issues arising from some like Ben Y. Zhao, who appears to have suffered a severe case of sour grapes.
Which has prompted Nicholas to write about his vulnerability research in,
“Why I attack”
https://nicholas.carlini.com/writing/2024/why-i-attack.html
It’s an interesting read in it’s totality but search down for,
“Unfortunately, the approach that Ben is taking is fundamentally flawed.”
Yup and it’s been known to be so since the end of the last century.
It highlights yet again a point I keep making about vulnerabilities
“The ICT Industry does not learn from it’s mistakes easily within living memory.”
If you want to know more about why the path Ben Y. Zhao is fundamentally flawed you need to go back and look at the work of Ross j. Anderson et al on the destruction of “Digital Watermarking” that was once the great hope of those desperate for DRM.
To see more specifically as to why look in Nicholas’s article for,
“Ben’s idea is that artists should add some adversarial noise to their images before releasing them to the public. This adversarial noise is supposed to make it so that, if someone tries to train a machine learning model on these images, the model will be bad.”
This is exactly what the tried to do with “Digital Watermarking” the only difference was that the “adversarial noise” in DRM was picked up by a chip rather than trying to “poison an LLM” etc AI system.
The underlying process is the same and is a spin off of “Low Probability of Intercept”(LPI) radio systems using “Direct Sequence Spread Spectrum”(DSSS) systems to send a semi-cocert signal with low visibility to an observer using conventional systems. It’s interesting to note that Military and Government radio systems have been moving away from this simple form of LPI since the 1990’s.
But there is another reason not to use Prof. Ben Y. Zhao’s and similar systems. As many readers here will know “adding noise to a signal degrades it significantly. For those that are not aware of how badly you can look up SINAD and how it effects the usability of a signal.
JonKnowsNothing • June 26, 2024 3:49 PM
@All
re: Influence Peddling P2
A timely court ruling from the US Supreme Court whether a payment made after the fact was a bribe or a gratuity.
Supreme Court wipes out anti-corruption law that bars officials from taking gifts for past favors
The Supreme Court on Wednesday struck down part of a federal anti-corruption law that makes it a crime for state and local officials to take gifts valued at more than $5,000 from a donor who had previously been awarded lucrative contracts or other government benefits thanks to the efforts of the official.
… the justices drew a distinction between bribery, which requires proof of an illegal deal, and a gratuity that can be a gift or a reward for a past favor. They said the officials may be charged and prosecuted for bribery, but not for simply taking money for past favors if there was no proof of an illicit deal.
The high court has long held that criminal laws restricting “illegal gratuities” to federal officials require proof that the gifts were given for a specific “official act,” not just because of the official’s position.
It may walk like a duck, quack like a duck, have feathers like a duck, have bill like a duck, have webbed feet like a duck, swim like a duck but it is not a duck in the view point of SCOTUS.
note: I will not include the source links, since moderation can be problematic. You will have to follow the breadcrumbs on your own.
Clive Robinson • June 27, 2024 4:04 AM
@ JonKnowsNothing,
Re : HP v. Autonomy
The reason it happened was that the HP board was highly toxic as a certain lady who shall remain nameless was spying on just about every senior decision maker, at the very least unlawfully.
As the driving person behind the deal within HP would not accept any argument or advice, the likes of due diligence did not happen at the level it should have, or if it did it was ignored.
Thus the acquisition became a “political rod” via shareholder action for somebody’s back and they in turn amped up the “throwing the toys out the pram” “way beyond eleven”.
The thing is HP got what they payed for, it was only after the “open book deal” was finalised check cleared and finalised and share holder type people started asking questions of the HP leaders about the money paid being excessive that trouble started.
You can read a timeline that is inaccurate in some of the early parts,
‘https://www.cio.com/article/304397/the-hp-autonomy-lawsuit-timeline-of-an-ma-disaster.html
And you can see what the game really was by looking through,
‘https://www.thestack.technology/autonomy-judgment-takeaways/
Put simply HP wanted the acquisition for “marketing reasons” that they figured would make HP be worth 30billion dollars. They cared not a jot about what the value really was it was all about “image” and “style” not actual worth.
It has been noted that HP’s “Due diligence” was a few phone calls over about six hours, and that later after the deal HP were well aware and did not care about financial irregularities (something that HP it’s self has carried out on more than one occasion).
What caused the issues was actually other areas of HP’s management. They ended up having to make a claimed 8billion write down so rather than have the truth revealed they then blamed not just the acquisition but previous HP management to try and save their own skins.
To this effect they claimed “fraud” but had no actual evidence and had made no formal complaint to any authority. Thus they had to make it up as they went along.
HP had not a snowball in hells chance in US Courts so they took a very weak scraped together case infront of a UK Judge in a civil not criminal court.
The judge after much prolonged “sitting on it” finally produced a massive report that basically said yes there had been some financial irregularities, but no they were in no way what HP was trying to claim and they had no chance of getting what they wanted.
In effect with the time consideration and the money HP Management threw in they have just been playing a delaying game to save their own skins and there will be no profit from the actions.
The behaviour of HP management is very reminiscent of the behaviour of the UK Post Office and we can only hope the nonsense will unroll faster than it did for the Post Office Scandal.
https://www.computerweekly.com/news/366588214/HP-loses-in-US-fraud-case-against-Autonomys-Mike-Lynch
In essence HP had to sue it’s self and admit it did wrong then use this admission to go after others…
It’s known as a “dog leg” and it’s quite dubious not just in use but by simple logic,
‘https://www.hausfeld.com/en-gb/what-we-think/perspectives-blogs/hp-v-autonomy-litigation-a-4-5-billion-dog-leg/
Oh and those HP shareholders… What have they got out of it? Well ~100million and a substantial loss in share value. As for HP management, well what do you think?
JonKnowsNothing • June 27, 2024 11:15 AM
@Clive, All
re: HP v. Autonomy | HP & Compaq
An earlier HP Boardroom scrap was the huge proxy fight over the Compaq acquisition.
iirc(badly) The board was divided into 2 factions: the Old-School and the Hayek-School. Deutsche Bank ended up being the Queen Maker and the acquisition was a colossal debacle.
iirc(badly) The folks at the top of that food chain are now prominent politicians and hang out with the movers and shakers in the Big Tech Feeding Trough.
Such deals are not uncommon and although the basic motto remains to increase Stockholder Maximum Value there is a significant benefit to others by holding enormous amounts of pledged debt. Pledged debt generates revenue for a different set of stockholders.
This latter is the basis of the Global Banking Industry, a different set of stockholders looking to Maximize Value but not overly worried about the debt holders. Current global estimate is that $1 of GDP generates $50 of Pledged Debt.
When GDP falls, no one takes a Hair Cut on Pledged Debt. We get Austerity instead. The desperate attempt to make debt payments on forever debt.
Subscribe to comments on this entry
Leave a comment
All comments are now being held for moderation. For details, see this blog post.
Sidebar photo of Bruce Schneier by Joe MacInnis.
Bruce Schneier • June 24, 2024 7:04 AM
All comments are now being held for moderation. For details, see this blog post.