SJC “Likes” Aspects of Facebook’s Work Product Argument


by Bret Cohen, Jillian Hart, and Matthew Brown

Case Focus

The Supreme Judicial Court (“SJC”) recently issued its anticipated decision in Attorney General v. Facebook, Inc., 487 Mass. 109 (2021), addressing the extent to which the work product doctrine or the attorney-client privilege protects internal investigations from disclosure.

The decision affirmed in part and reversed in part a Superior Court decision (Attorney General v. Facebook, Inc., 2020 WL 742136 (Jan. 17, 2020) (Davis, J.)) that held that the work product doctrine did not apply to documents the Massachusetts Attorney General (“AG”) sought from social media giant Facebook, Inc.’s (“Facebook”) internal investigation into a data privacy breach.

The SJC’s decision provides an important reminder for companies to tread carefully and always consult with counsel before launching an internal investigation to understand what may be discoverable in future litigation. 

Facts and Background

After a widely publicized data breach incident involving one of its third-party applications (“apps”), Facebook undertook an internal investigation, led by outside counsel, to determine the extent to which the platform’s apps misused user data and to evaluate associated liability.  Facebook intended for the app developer investigation (“ADI”) to identify any other apps that misused user data and to assess Facebook’s potential liability from the incident.  Both in-house and outside counsel “designed, managed, and overs[aw]” the ADI and “devised and tailored the ADI’s methods, protocols, and strategies to address the specific risks posed by these legal challenges.”  Outside counsel also retained third-party technical experts and investigators to assist in the ADI.  

As a result of the data breach incident, the AG opened its own investigation into whether Facebook misrepresented the extent to which it protected or misused user data.  In accordance with its authority under M.G. L. c. 93A, the AG issued a series of civil investigative demands to Facebook.  Facebook complied in part, but refused on privilege grounds to honor six of the AG’s requests.  The first five requests sought information related to the identities of certain apps and app developers that Facebook identified and reviewed during its ADI.  The sixth request, in contrast, sought Facebook’s internal communications and correspondence regarding certain apps.

The AG filed a petition to compel compliance with its demands in the Superior Court’s Business Litigation Session.  The Superior Court sided with the AG, holding that the work product doctrine did not cover Facebook’s ADI and, even if it did, the AG made the required showing of a substantial need for the information that it could not obtain without undue hardship.  As to Facebook’s asserted attorney-client privilege, the Superior Court held that the privilege did not cover the information sought by the AG’s first five requests, and ordered the production of responsive documents.  Regarding the sixth request, however, the Superior Court held that it did seek the disclosure of potentially privileged material, and accordingly ordered Facebook to provide the AG with a detailed privilege log.

Facebook appealed and the SJC heard the case on direct appellate review.


The SJC affirmed in part and reversed in part the Superior Court’s decision. 

Work Product Doctrine:  Addressing the first five requests, the SJC held that the work product doctrine did apply, because Facebook conducted the ADI in anticipation of litigation.  The Court, however, carefully distinguished fact work product from opinion work product.  It held that as to documents constituting fact work product, the AG demonstrated substantial need and undue hardship requiring production.  At the same time, the SJC held that remand was necessary to determine whether any of the responsive documents that Facebook withheld constituted opinion work product.  The SJC held that, if any of the withheld documents constituted opinion work product, such documents are only “discoverable, if at all, in rare or extremely unusual circumstances.”  Facebook, 487 Mass. at 128 (internal quotations omitted).

Attorney Client Privilege: Addressing the sixth request, the SJC agreed with the Superior Court that to the extent Facebook objected on the basis of attorney-client privilege, Facebook must produce a detailed privilege log so that the AG could assess (and potentially contest) the privilege assertions.  The SJC reasoned that the request sought information dating back years before the ADI began, as well as communications that did not involve attorneys, both of which might fall outside the scope of the attorney-client privilege.  The SJC held that the attorney-client privilege did not cover the first five requests as they merely sought underlying facts rather than attorney-client communications.  In doing so, the SJC emphasized that “the attorney-client privilege only protects communications between attorneys and a client about factual information, not the facts themselves,” noting that “this distinction is important and somewhat collapsed by the advocacy in the instant case.” Facebook, 487 Mass. at 123.

Key Issues

In its decision, the SJC identified and discussed three important issues pertaining to employers contemplating or conducting internal investigations.

First, the SJC considered whether the work product doctrine applies to an internal investigation.  In the instant case, the SJC held that the work product doctrine applied to the ADI because: (1) documents were prepared; (2) by or for Facebook for its agents; and (3) in anticipation of litigation.  The SJC specifically found that, although Facebook had an ongoing compliance program, the ADI was “meaningfully distinct” from the compliance program, with its own distinct methodology focused on past violations, rather than improving ongoing operations in the normal course of business.  In short, the mere fact that the ADI also served Facebook’s business purposes did not mean that the work product doctrine was inapplicable.

Second, the SJC discussed whether the information sought by the AG constituted fact work product or opinion work product.  As the SJC stated, “the line between fact work product and opinion work product is not always clear.”  In this regard, the SJC noted that although Facebook made multiple public statements about the ADI and the investigatory process (which Facebook could not then claim to constitute opinion work product), any “undisclosed strategic decision-making by counsel, including the assessment of legal risk or liability [] revealed by the factual analysis” might qualify as opinion work product.    

Third, the SJC considered whether, in relation to fact work product, the party seeking disclosure established a substantial need for and undue hardship from denied access to the work product sufficient to warrant its discovery.  The SJC held that the AG met its burden by demonstrating both.  With respect to the AG’s substantial need, the SJC found that the app-related information sought was central to the statutorily authorized c. 93A investigation.  Likewise, with respect to the AG’s asserted undue hardship, the SJC distinguished the ADI from a routine internal investigation that “involved simply interviewing key employees and other witnesses or reviewing a manageable number of documents, tasks that can be easily replicated by third parties or government investigators.”  Here, the ADI was a years-long investigation involving a vast quantity of information and included analysis of millions of apps by hundreds of outside experts. Therefore, the SJC ruled that the enormous costs and time required to duplicate the ADI was sufficient to demonstrate undue hardship.


Although this area of law is far from settled, the Facebook decision provides helpful guidance for companies contemplating and conducting internal investigations.  Key guideposts include:

  • Engage counsel in advance of an internal investigation to discuss the objective and parameters of such investigation.
  • Any outside experts involved in the investigation should be retained by outside counsel and should be bound by confidentiality agreements.
  • Review what records and files the company develops in the regular course of business and be mindful that these records may be discoverable if not created in anticipation of litigation.

During the internal investigation, consider carefully what information and documents may be characterized as fact (versus opinion) work product and, therefore, may be discoverable.  As the SJC cautions, the line between the two is “not always clear” and, consequently, aspects of internal investigations, especially fact work product, may be discoverable.

Bret Cohen chairs the Labor & Employment and Trade Secrets & Employee Mobility Practice Groups at Nelson Mullins Riley & Scarborough LLP. His practice covers a wide range of areas, including the enforcement of non-compete and employment agreements, complex commercial and trade secrets litigation, and advice and counsel on termination and transition of high-level executives.  

Jillian Hart is an associate in the Labor & Employment Group at Nelson Mullins Riley & Scarborough LLP. Jillian focuses her practice on employment and trade secrets litigation and also advises clients on a variety of employment matters, including restrictive covenants and wage and hour issues.   

Matthew Brown is an associate in the Labor & Employment Group at Nelson Mullins Riley & Scarborough LLP. Matthew focuses his practice on trade secrets and non-compete litigation and advice and counsel on a variety of issues, including worker classification and employment agreements. 

Making “Good” Laws Through the Ballot Initiative Process

Kim_Tori by Tori T. Kim

Practice Tips

From marijuana legalization to campaign finance reform to a constitutional amendment to impose a “millionaire’s tax,” citizen groups turned to the initiative petition process this year to propose a variety of public policy measures. The process, governed by article 48 of the amendments to the state constitution, allows citizens to place measures directly on the ballot as an alternative to enacting legislation through elected representatives. Twenty-three other states permit similar forms of “direct democracy.” But compared to some systems (notably, the much-criticized California model), the Massachusetts process contains comparatively strict requirements to help ensure public support before a measure reaches the ballot and to make better law.

The initiative petition process is straightforward in theory but complex in its implementation. It begins with a filing with the Attorney General’s Office, usually by the first Wednesday in August of the year preceding a biennial state election. If the petition is “certified” by the Attorney General, the petitioners must then collect thousands of signatures by the first Wednesday in December in order to present the petition to the legislature. The legislature can choose to enact the petition in the same form or take no action by the following May, and, in the latter event, the petitioners must gather more signatures in order to place the petition on the November ballot. Proposed constitutional amendments follow a similar process, except that the measure must receive at least 25 percent support in joint sessions of two successive legislatures before it can appear on the ballot. Thus, a proposed constitutional amendment submitted in 2015 could not appear on the ballot until the 2018 election year.

Article 48 also restricts the types of initiative petitions that may appear on the ballot. Among the most litigated limitations is the requirement that the petition must contain “only subjects . . . which are related or [] mutually dependent.” Art. 48, The Initiative, II, § 3. In Carney v. Attorney General, 447 Mass. 218 (2006), the Supreme Judicial Court construed this phrase narrowly as requiring that a measure reflect an “operational relatedness among its substantive parts that would permit a reasonable voter to affirm or reject the entire petition as a unified statement of public policy.” Id. at 230-31. The Court applied this standard to deny certification of a petition seeking simultaneously to ban the dog racing industry and to increase penalties for the inhumane treatment of dogs. Although the Carney standard did not pose a hurdle for this year’s petition to legalize marijuana for adult users, similar petitions that address one “subject” broadly, but seek to make reforms in many “operationally” unrelated areas of the law, could be susceptible to challenge.

Importantly, article 48 bars petitions that are “inconsistent” with certain rights enumerated in the Declaration of Rights. See art. 48, The Initiative, II, § 2. For instance, the Supreme Judicial Court in Bowe v. Secretary of the Commonwealth, 320 Mass. 230 (1946), denied certification of a petition proposing to eliminate all forms of political spending by labor unions as “inconsistent” with unions’ free speech and assembly rights. Id. at 252. However, the list of rights in article 48 is limited, reflecting a compromise among the members of the constitutional convention to prospectively allow voters to “override” decisions of the state’s highest court only in certain areas. The members specifically had in mind Lochner-era cases declaring social welfare legislation invalid as violating “due process” as a type that could be addressed by an initiative petition, but they identified other “concrete” and “definite” rights enumerated in the Declaration of Rights that would not be subject to the initiative petition process. This compromise impacts advocates of all political persuasions, as is evident from this year’s petition to roll back corporate political spending in a manner similar to that in the Bowe petition banning labor union spending.

In addition, article 48 bars initiative petitions that make a “specific appropriation of money from the treasury of the commonwealth.” While this limitation preserves the legislature’s exclusive authority to make appropriations, it does not prohibit a petition from specifying how funds may be spent once they are so appropriated. For instance, this year’s petition imposing an additional 4% tax on incomes over $1 million states that the revenues collected under this provision shall be spent for the purposes of enhancing public education and transportation, but specifies that such spending is “subject to appropriation.” While this could mean that the legislature may decline to appropriate the collected revenues for the stated purposes, the fact that the “millionaire’s tax” is proposed as a constitutional amendment—which requires at least 25 percent support of the legislature—could reduce the chance of such a result. So too may the legislature’s separate duty under article 48 to “appropriate such money as may be necessary to carry such law [if passed] into effect.” Art. 48, The Initiative, II, § 2; see also Bates v. Director of the Office of Campaign and Political Finance, 436 Mass. 144, 154-61 (2002).

Whatever one’s views on the effectiveness of the initiative petition process as a means of making public policy, everyone should agree that any measure that is destined to become law should be well-drafted. A few suggested guidelines in this regard include the following:
• Research the law to ensure consistency with existing provisions. Some changes proposed by a petition could be achieved through existing law or a more modest modification of such law.

• Consider the impact of the petition on other areas of the law. For instance, a change in the definition of a term could affect every provision of the General Laws where that term is used.

• Keep the legislative language succinct. An often-cited rule of thumb is to draft a summary of the petition as it would appear on the ballot, and then craft legislative language to match the summary.

• Consider issues that may subject the law to constitutional or other challenges if the petition were enacted, even if such issues would not bar certification. For instance, laws that have retroactive effect could raise due process issues.

These suggestions could help reduce duplication and confusion in the law, while also keeping issues succinct and clear for the voters. Overall, they further the goal of making “good” workable laws, in accordance with the overriding purpose of article 48.

Tori T. Kim is Deputy General Counsel in the Executive Office for Administration and Finance. Previously, as Assistant Attorney General, she co-directed the review of initiative petitions at the Attorney General’s Office.